ODEON

Basics Industrial Auditorium Combined

User’s Manual
ODEON Room Acoustics Software
User’s Manual
Version 17

Published in December
2021

Odeon A/S
DTU Science Park
Diplomvej, building 381
DK-2800 Kgs. Lyngby
Denmark
Tel: +45 8870 8845
Fax: +45 8870 8090
www.odeon.dk

0
END USER LICENSE AGREEMENT
This legal document constitutes the complete agreement between you, the end user (either an
individual or a single entity) and Odeon A/S. All rights not expressly granted in this License
Agreement are reserved by Odeon A/S.

The right to use the ODEON software is provided only on the condition that the user agrees to the
following License Agreement.

COPYRIGHT/PROPRIETARY PROTECTION
The ODEON software, auxiliary applications, and documentation are owned by Odeon A/S and
are protected by copyright laws and international copyright treaties, as well as other intellectual
property laws and treaties. The SOFTWARE PRODUCT is licensed, not sold.

You shall be held legally responsible for any copyright infringement that is caused or encouraged
by your failure to abide by the terms of this agreement.

OWNERSHIP OF ODEON
Odeon A/S retains sole title and ownership of the ODEON software, regardless of the form or
media in or on which the original and other copies may exist. As the licensee, you own the storage
physical media on which the ODEON program is subsequently recorded or fixed. This Agreement
is not a sale of the original ODEON software or any copy thereof. All rights are reserved by Odeon
A/S.

GRANT OF LICENSE
In consideration of payment of the License fee, Odeon A/S grants each customer (a private person
or a company registered in a country) a non-exclusive right to use the ODEON software. This
agreement grants you the right to make copies of the software for archival purposes (eg. on a hard-
disk) The software may be loaded onto multiple computers, but can only be fully operational under
any of the following licensing schemes:
1. A hardware key (dongle) in form of a First or Extra license is attached via USB on the same
computer that operates ODEON. Remote connections are not possible and not allowed in this
case.
2. If the "Windows Remote desktop" option has been purchased with a First or Extra license, then it
is possible to run the ODEON software remotely. In this case, the hardware key should be
attached on a remote server/desktop computer, where the software is installed. Any other
computer can get access to the server/desktop computer and operate ODEON remotely.

HARDWARE KEY
A license for using the ODEON software is granted via a hardware key. The hardware key holds the
license information and determines the version and edition of the ODEON software that can run
when the hardware key is attached to the computer. If a hardware key is defective during the
warranty period (first licensed year), Odeon A/S will replace it. The customer will pay a fee for
replacing the hardware key, should it become defective after the warranty period. The hardware
key contains the name and country of the licensee which are displayed inside the software, on print
outs from the software and on graphics exchanged from the software. The ODEON Combined Free
trial version of ODEON has limited functionality and runs without a key.

1
LICENSE RESTRICTIONS
Modification, adaptation, translation, reverse engineering, de-compiling, disassembling, attempts
to discover the source code of the software, or the creation of derivative work based on ODEON
software are strictly prohibited.

In no way may you transfer, assign, rent sublicenses, lease, sell, or otherwise dispose of any portion
of the software on a temporary or permanent basis.

TRANSFER RESTRICTIONS
The ODEON software is licensed only to the licensee. In no way may you transfer, assign, rent
sublicenses, lease, sell, or otherwise transfer any portion of the license on a temporary or permanent
basis (this includes transfer between branches of international companies that are registered in
different countries). The license may only be transferred to anyone if Odeon A/S accepts (e.g. in the
case of merging two companies) and with the prior written consent of Odeon A/S. Any authorized
transfer of the ODEON software shall be bound by the terms and conditions of this Agreement.

TERMINATION
This license is effective until termination. This license will terminate automatically without notice
from Odeon A/S if you fail to comply with any provision of this Agreement.

Upon termination, you shall destroy the written materials and all copies of the ODEON software,
including archival copies, if any.

NO LIABILITY FOR DAMAGES

Neither Odeon A/S nor anyone else who has been involved in the creation, production, distribution
or delivery of this program shall be liable for any direct, indirect, consequential, or incidental
damages (including, without limitation, damages for loss of profits, business interruption, loss of
information, incorrect results, recovery of data, or other pecuniary loss) resulting from any defect
in the software, update version, or its documentation, or arising out of the use, the results of use,
or inability to use the product, even if Odeon has been advised of the possibility of such damage
or claim.

LIMITED WARRANTY
Odeon A/S makes no warranty or representation, either expressed or implied, including, but not
limited to, implied warranties or merchantability and fitness for a particular purpose, with respect
to this product. As a result, this software is sold "as is" and you the licensee are assuming the entire
risk as to its quality and performance. No Odeon A/S Distributor, agent, or employee is authorized
to make any modification, extension or addition that will increase the scope of this warranty.

UPDATE POLICY
Odeon A/S may create, from time to time, updated versions of the ODEON software. Odeon A/S
reserves the right to make changes to the software or its documentation without notice. Major
updates e.g. from ODEON 16 to ODEON 17 will be made available to registered users who have a
valid Maintenance and Support Agreement or who purchase an upgrade. Minor service updates
e.g. from version 17.0 to 17.01 or 17.1 may be made available for download and may also be used
by the licensee without a valid Maintenance and Support Agreement if the hardware key permits
running the update.

2
USE OF ANECHOIC ORCHESTRA RECORDINGS DELIVERED WITH
THE SOFTWARE
The ODEON Auditorium and Combined versions (excluding the demo version) install with a
number of anechoic recordings that may be used with the ODEON software without restrictions.
Two separate installation packages, OdeonWavePack_I and OdeonWavePack_II, are available on
the installation USB stick. These WavePacks install anechoic recordings of orchestra pieces along
with room examples demonstrating the recordings.

For the orchestra recordings in OdeonWavePack_I special restrictions apply:

• Customer acknowledges that the anechoic recordings are the sole property of TKK,
Helsinki University of Technology.
• Customer acknowledges the right to use the recordings for internal use only and the right
to make one copy for archival and backup purposes.
• Customer agrees not to modify or remove, and not to tamper with or disable any
automated display, performance, or execution of, any credits, references or notices,
without the prior written consent of TKK, Helsinki University of Technology.
• Customer agrees not to copy, modify, adapt, combine, publish, rent, lease, sell, distribute
the recordings or sublicense or otherwise transfer the sublicense.
• If the anechoic recordings are used in the Customer's work in such a way that an ordinary,
reasonable user familiar with the recordings would be likely to recognize it, Customer
agrees to include the following notice in the resulting work: "Derived from anechoic
symphony orchestra recordings by Jukka Pätynen and Tapio Lokki, Helsinki University of
Technology and licensed from Odeon A/S".

For the orchestra recordings in OdeonWavePack_II the following restrictions apply:

• Customer acknowledges that the anechoic recordings are the sole property of Odeon A/S,
Denmark.
• Customer acknowledges the right to use the recordings for internal use only and the right
to make one copy for archival and backup purposes.
• Customer agrees not to modify or remove, and not to tamper with or disable any
automated display, performance, or execution of, any credits, references or notices,
without the prior written consent of Odeon A/S.
• Customer agrees not to copy, modify, adapt, combine, publish, rent, lease, sell, distribute
the recordings or sublicense or otherwise transfer the sublicense.
• If the anechoic recordings are used in the Customer's work in such a way that an ordinary,
reasonable user familiar with the recordings would be likely to recognize it, Customer
agrees to include the following notice in the resulting work: "Derived from anechoic
symphony orchestra recordings at the Technical University of Denmark and licensed from
Odeon A/S".
Additional anechoic recordings are available from the ODEON homepage at:
https://odeon.dk/downloads/anechoic-recordings/

GENERAL
This agreement constitutes the entire agreement between you and Odeon A/S concerning the
ODEON product. If any provision of this Agreement shall be declared void or unenforceable, the
validity of this Agreement and all other provision shall not be affected. ODEON product support
and service is only provided to registered ODEON users with a valid support agreement.

3
Contents
END USER LICENSE AGREEMENT ...................................................... 1
Contents......................................................................................................... 4
Introduction .................................................................................................. 7
ODEON Installations .............................................................................. 9
1.1 Installing and running the program ................................................................................................ 9
1.2 Upgrading from previous versions .................................................................................................. 9
1.3 What’s new in ODEON 17 .............................................................................................................. 11
1.4 Features of previous versions ......................................................................................................... 19
1.5 How to upgrade or update your current license.......................................................................... 21
Modelling Rooms ................................................................................... 24
2.1 Pre-calculated Rooms....................................................................................................................... 24
2.2 Guidelines on room modelling ....................................................................................................... 27
2.3 Files created by ODEON ................................................................................................................. 29
2.4 Importing Geometry into ODEON ................................................................................................ 30
2.5 Importing a model from SketchUp ................................................................................................ 31
2.6 Importing CAD files ......................................................................................................................... 34
2.7 The ODEON Extrusion Modeller ................................................................................................... 40
2.8 Model check in ODEON .................................................................................................................. 46
2.9 Combining geometries ..................................................................................................................... 49
Sources and Receivers ........................................................................... 50
3.1 Generic point sources ....................................................................................................................... 50
3.2 Natural point sources....................................................................................................................... 50
3.3 Common loudspeaker format, CF1 and CF2 files ........................................................................ 52
3.4 Directivity files in the EASE 3 and EASE 4 formats..................................................................... 52
3.5 Creating directivity patterns in ODEON’s .So8 format ............................................................... 53
Materials................................................................................................... 57
4.1 Overview of the Material list .......................................................................................................... 57
4.2 Categories in the Material library .................................................................................................. 60
4.3 Editing/extending the Material library .......................................................................................... 61
4.4 Editing the Surface list ..................................................................................................................... 63
4.5 Material calculator ............................................................................................................................ 65
4.6 Scattering coefficient ........................................................................................................................ 71
4.7 Transparency coefficient – semi-transparent surfaces ................................................................ 74
4.8 Different Wall Types ........................................................................................................................ 75
4.9 Material Toolbar ............................................................................................................................... 75
Calculation results.................................................................................. 77
5.1 Global Calculations .......................................................................................................................... 77
5.2 Point response calculations ............................................................................................................. 78
5.3 3D-matrix Calculation...................................................................................................................... 81
5.4 Results from measurements ............................................................................................................ 82
5.5 Visualisation Tools ........................................................................................................................... 82
Colour Grids – colour maps ................................................................. 83
6.1 Overview ........................................................................................................................................... 83
6.2 Accuracy in the Interpolated Grids ................................................................................................ 85

4
6.3 Range, Colours and Labels for Grid Display ................................................................................ 86
Auralisation ............................................................................................. 87
7.1 What is Auralisation ........................................................................................................................ 87
7.2 Auralisation Options ........................................................................................................................ 89
7.3 Setup for Auralisations .................................................................................................................... 89
7.4 Making auralisations........................................................................................................................ 94
7.5 Audio Effects – Multiplying the Audio Files .............................................................................. 100
7.6 Examples of Auralisation .............................................................................................................. 105
7.7 Sound Pressure Levels for Auralisations .................................................................................... 109
7.8 Input/Output Signals for Auralisations ....................................................................................... 110
Room Acoustical Parameters.............................................................. 112
8.1 Decay curve ..................................................................................................................................... 112
8.2 Reverberation Time and Early Decay Time ................................................................................ 113
8.3 SPL, G, C80, D50, LF80, STearly, STlate and STtotal ............................................................................... 122
8.4 Stage Parameters ............................................................................................................................ 124
8.5 IACC – Inter-Aural Cross Correlation ......................................................................................... 125
8.6 STI – Speech Transmission Index ................................................................................................. 126
8.7 DL2 – Rate of Spatial Decay ........................................................................................................... 127
8.8 ISO 3382-3 (Open-plan offices) parameters ................................................................................ 128
8.9 Edit/Make Room Acoustic Parameters ........................................................................................ 129
Calculation Principles ......................................................................... 131
9.1 What can be modelled in ODEON ............................................................................................... 131
9.2 Global decay .................................................................................................................................... 132
9.3 Response from Sources to Receivers ............................................................................................ 133
9.4 Angle-dependent absorption ........................................................................................................ 140
9.5 Scattering ......................................................................................................................................... 142
9.6 Oblique Lambert ............................................................................................................................. 147
9.7 Diffraction over screens and round objects (Screen diffraction) .............................................. 149
9.8 Radiated rays from a source ......................................................................................................... 150
9.9 Processing reflection data for auralisation in Single Point Response...................................... 151
9.10 Calculation method for Reflector Coverage ............................................................................. 151
Calculation Parameters ..................................................................... 152
10.1 Impulse response length (Job calculations and Global Estimate) .......................................... 153
10.2 Number of late rays (Job calculations) ...................................................................................... 153
10.3 Specialist settings .......................................................................................................................... 155
Achieving Good Results ................................................................... 158
11.1 What are Good Results ................................................................................................................ 158
11.2 Sources of Error ............................................................................................................................ 159
Measurement System ........................................................................ 162
12.1 Technical Background ................................................................................................................. 164
12.2 Equipment to Use with the Measurement System ................................................................... 165
12.3 Measurement Setup Options ...................................................................................................... 169
12.4 Load Impulse Response ............................................................................................................... 174
12.5 Results from Measurements ....................................................................................................... 176
12.6 Inserting measured room acoustics parameters into a multi-point response ...................... 178
12.7 Editing Impulse Responses ......................................................................................................... 179
12.8 Making Auralisations with Impulse Responses ....................................................................... 180
12.9 Calibration of Measurements...................................................................................................... 181
12.10 Examples of Impulse Responses – Warnings ......................................................................... 189

5
 Genetic Material Optimizer ............................................................. 198
13.1 General Steps for Material Calibration ...................................................................................... 198
13.2 How Genetic Algorithms Work.................................................................................................. 201
13.3 Perform a Material Optimization ............................................................................................... 203
Source Power Estimation .................................................................. 212
14.1 Mathematical background .......................................................................................................... 212
14.2 Perform a source power estimation ........................................................................................... 213
14.3 Analysing the results ................................................................................................................... 214
14.4 Summary of the general steps .................................................................................................... 216
14.5 Best practice guide to optimise your results ............................................................................. 216
Line Array Sources ............................................................................. 217
15.1 Stacking the units ......................................................................................................................... 217
15.2 Playing with delay........................................................................................................................ 218
15.3 Playing with level ......................................................................................................................... 220
15.4 Combining delay and level adjustments ................................................................................... 221
15.5 Using the equalizer ...................................................................................................................... 221
15.6 Bringing the array into the room ................................................................................................ 221
References ................................................................................................. 223
Appendix A: Vocabulary ....................................................................... 227
Appendix B: Specify Transmission Though Walls .......................... 230
Appendix C: Importing Array Loudspeaker Data (XML format) .. 233
Appendix D: Inverse filter design ....................................................... 243
Calculation principles .......................................................................................................................... 243
Interface of the inverse filter tool ....................................................................................................... 245
Remarks on multichannel files ........................................................................................................... 246
Appendix E: Head Related Transfer Functions (HRTF’s) and digital
filtering ...................................................................................................... 247
Appendix F: Impulse Response processing tools ............................. 250
Create average of multiple impulse responses (.wav) .................................................................... 250
Create impulse response file (.wav) from spectrum (.txt) .............................................................. 251
Create inverse filter (.wav) .................................................................................................................. 253
Typical cases for using the tools above ............................................................................................. 254
Appendix G: Modelling rooms in the ODEON Editor .................... 256
Appendix H: Mathematical Expressions ............................................ 288

6
Introduction
Thank you for using ODEON, the Room Acoustic Software package that helps you to make
complete studies of room acoustics for any large space through simulations and measurements!
We hope you will enjoy working with ODEON and that you will find it extremely useful in the
acoustical design of new spaces and improvement of existing ones. ODEON has a very user-
friendly interface, which allows setting up a project very fast, and to create elegant presentation of
results.

This manual is intended to serve as an introduction on modelling room geometries in the ODEON
software, on the features for measurements and simulations, and on the applied calculation
principles. It will not cover all features included in the ODEON software in detail; explanations of
displays, calculation parameters, results, etc. are available as context sensitive help from within the
ODEON application (shortcut F1). It is recommended to use the context sensitive help to learn about
the specific features available from the different displays, the interpretation of results, calculation
parameters, etc.

A large number of comprehensive tutorial videos are also available from our home page – they
can be accessed directly from within ODEON using the Shift+Ctrl+F1 shortcut or by clicking the
Video tutorial icon in the ODEON toolbar.

The contents of this manual are as follows:

Chapter 1 covers installation of the program, changes from previous versions, etc.

Chapter 2 covers geometry modelling, with important tips on how to make models for acoustical
simulations. It also covers how to import geometry from 3D modelling software, and how to use
the 3D modelling tools in ODEON.

Chapter 3 describes types of sources and receivers, how to extend the library of source directivity
patterns available for point sources, and the use of directivity patterns in the Common
Loudspeaker Format (CLF).

Chapter 4 deals with the acoustic materials to assign to the surfaces of rooms; absorption, scattering
and transparency coefficients, as well as transmission data. Special materials that may speed up
the modelling process, and how to manage the material library, are also covered in this chapter.

Chapter 5 deals with different calculation options and associated results in the four editions of
ODEON.

Chapter 6 describes how to work with Colour Grids.

Chapter 7 deals with the auralisation options in ODEON Auditorium and Combined; the hardware
needed for auralisation, how to publish calculated sound examples, etc.

Chapter 8 describes important calculated room acoustic parameters available in ODEON, how they
are calculated, and how to interpret results.

Chapter 9 introduces the calculation principles used in ODEON, giving an idea on the capabilities
and limitations of the simulations performed by the ODEON software.

7
Chapter 10 describes the various calculation parameters available in the program. Most of the
parameters are automatically set to reasonable values by ODEON; However, for special cases you
may need to adjust some of the calculation parameters.

Chapter 11 discusses quality of results and how to achieve good results. This chapter may be
relevant once familiar with the program.

Chapter 12 presents the integrated impulse response measuring system in ODEON. The chapter
explains which measuring methods are used, how to obtain a healthy impulse response and how
to obtain room acoustic parameters as the results. The chapter ends with an annex giving
examples of good and bad impulse responses.

Chapter 13 describes how to use ODEON’s Genetic Material Optimizer for calibrating a room model
towards a real room. The Genetic Material Optimizer makes use of genetic algorithms in order to calibrate
the materials (absorption coefficients) in the simulation model for good agreement between
simulated and measured room acoustics parameters.

Chapter 14 presents ODEON’s Source Power Estimation tool which makes it possible to estimate the
source power spectrum of sources from real-life SPL measurements.

Chapter 15 introduces the line array sources discussing some basic principles behind array source
design.

8
 ODEON Installations
To work with ODEON you need both an installation of the program on your PC, as well as a
hardware key dongle (Rockey 6 Smart – R6 Smart). After purchasing ODEON, you receive the
dongle with time limitation. Once full payment is completed, an update file is sent by Odeon A/S
that removes the time limitation. The update file needs to be installed by you on the dongle.

1.1 Installing and running the program

To run ODEON, your PC must be running one of the operating systems supported:
• 32 bit/64 bit Windows 10.
• 32 bit/64 bit Windows 8.
• 32 bit/64 bit Windows 7.
• 32 bit/64 bit Windows Vista.

ODEON comes in a USB drive containing the edition of ODEON that was purchased (Basics,
Industrial, Auditorium or Combined). As an alternative, you may download the most recent updated
version, including service updates that were developed since the production of the USB, from
https://odeon.dk/downloads/odeon-installations-updates/. To install the program:

a) Double-click on the file with the name of the edition you wish to install, e.g.
InstallOdeon14Combined.exe to install the program.
b) To run the program, the supplied hardware key dongle (Rockey 6 Smart – R6 Smart) must be
inserted into the USB port on the PC. If you start the program without the hardware key, it
can only be used in viewer mode.
c) When you have installed the ODEON software, it will automatically check for updates. If there
are updates available, ODEON will update your installation if you accept.
d) Initially, the dongle ships with a time limitation. Once full payment is completed, send a
request to [email protected] to remove the time limitation. You will receive an update file that
needs to be installed on the dongle by choosing Tools>Download license update to dongle (.cif).

1.2 Upgrading from previous versions

If you are upgrading from previous versions of ODEON, read below to learn about the changes in
ODEON. Below is a list of issues which you should be aware of when upgrading from previous
versions.

Upgrading to version 17
ODEON version 17 runs on Windows 10 (32/64 bit editions) – ODEON 17 may be able to run on
Windows Vista and Windows 8, but has not been tested on these operating systems which are
unsafe and no longer maintained and supported by Microsoft. ODEON 17 is a major upgrade,
please see sec. 1.3. Though ODEON 17 is capable of loading and converting older projects into
version 17 format (forward compatibility), an older version of ODEON may not load a room once
it has been loaded into ODEON 17. ODEON 17 includes a large number of enhancements – for a
fairly complete list see sec. 1.3 or open the help file from within the ODEON software (press F1
from within ODEON, then click the What’s new in ODEON 17 entry in the contents-menu).

Remember: If keeping an earlier version, be careful not to mix the use of old and new versions –
although we do strive to maintain forward compatibility, we cannot guarantee that a room which has
been loaded into a newer version of ODEON will also load into an older version without problems.

9
Major upgrade
If performing a major upgrade, typically a full version number or more, e.g. from version 16 to
version 17, then ODEON will install to a new directory for that version without changing the
existing installation. If you have no wishes to use the old version of ODEON then it is
recommended to uninstall the version(s) using the Windows Start|Control Panel|Add/remove
programs feature.

If you have been using version 7 or earlier

If you are used to versions earlier than version 8, it is essential to learn about the new methods for
handling of scattering. Chapter 4 covers the material properties to assign to surfaces; Chapter 9
covers the calculation principles including handling of scattering and Chapter 10 covers the choice
of calculation parameters.

If you are having problems loading a room, which was created and worked fine in an old version
of ODEON, this is probably due to a change that has been made to the surface numbering
mechanism applied in ODEON. The numbering mechanism has been changed slightly in order to
avoid a conflict which appeared when using ‘symmetric surfaces’, along with modelling entities
such as CountSurf, Box, Cylinder etc. and in particular to make the automatic surface numbering
work without any problems (when the NumbOffSet is set to Auto).

If having problems loading a room due to the reasons just mentioned, ODEON will either give an
error message that surfaces are repeated in the geometry file or that materials are not applied to all
surfaces. In these cases you may wish that ODEON uses the old numbering mechanism – this can
be done using the Version4 .par file. At the first line in the geometry file, just after the ### sign, type:
flag in the Version4 TRUE.

Once the old incompatible code ends, Version4 may be set to FALSE again.

If you have been using a version 6 and earlier

If you are used to a version earlier than 6 then we do recommend that you read carefully through
the manual as if you are a newcomer to ODEON. There are substantial differences between the
early versions of ODEON and the ODEON software as it is today – modelling has been made easier,
calculation principles have been enhanced and a huge number of new features has been added.

Project files from very old versions of ODEON

The only project file from Odeon versions earlier than 3.0, being fully compatible, is the surface file
(.sur). The rest of the project files are no longer valid. And even though the .sur format is still valid,
it is not recommended to model rooms in this format. The .par format is a much more efficient
format.

10
1.3 What’s new in ODEON 17
The new version of ODEON is packed with attractive features and enhancements, listed below.
The headings will specify what editions of ODEON: (B) Basics, (I) Industrial, (A) Auditorium and
(C) Combined, each feature is available in. If not specified, then it is available for all editions.

Remember: The ODEON software is only forward compatible. This means that you may load a room
that has been created in ODEON 16 (or earlier versions down to ODEON 3.0) into ODEON 17. The
opposite is not always the case. Once a room has been loaded into ODEON 17, there might be issues
loading it into earlier versions. ODEON 17 will issue a warning before loading a room created in an
earlier version - allowing you to make a backup copy for use in an older version of ODEON.

Material Calculator (A)(C)

A Material calculator, similar to products such as Norflag and Zorba, is included in ODEON 17
Auditorium and Combined. This tool allows estimating absorption of materials, such as porous
absorbers (e.g. suspended ceilings), resonators (perforated as well or slit perforated) and thin
membrane materials. You can vary the thickness and configuration by adding more than one layer.
The tool will calculate the overall absorption coefficient of the compound structure. You can access
it from within the material list by clicking the Add material button or by editing an existing
calculated material from the list. The material calculator is the last TabSheet in the add or edit
window. See a comprehensive description in Chapter 4.

Improved Calculation algorithms

Calculation algorithms for simulation of impulse responses have been slightly improved for
coupled spaces. In cases where no early reflections (images sources) are found for a given source
at a given early reflection order, ODEON will revert to its late reflection method. As a consequence,
the default Transition Order (TO) has been increased from 2 to 3 for optimum results. Good
results are also obtained with higher TO, e.g. 4-5, though calculation times will increase. Quality of
results is improved when the sources are located in another cavity than the receiver - a good
example is the side galleries of Hagia Sofia (a room installed with ODEON in the
Odeon17Combined\Rooms\CAHRISMA Project) when the source is located at the altar.

11
 Cluster area algorithm
ODEON will calculate the joint area of surfaces which are in the same surface plane, have joining
edges and use the same material. This cluster area is used for angle depending absorption and for
calculation of the reflection-based scattering coefficient. ODEON will automatically recalculate
cluster areas whenever materials are being (re-)assigned. As a consequence, the Fractional type in
the Material List does not have to be specified by the user as often as before. A more precise area is
applied by Odeon now. For concave shapes, like cylinders and domes, the Fractional type must still
be used in order to estimate focusing. This setting may also be used for surfaces which have same
or similar materials and are located in almost same planes.

Let’s assume a "chessboard-like" surface with 3x7=21 tiles - with alternating hard reflecting and
soft absorbing areas. The cluster area will be "1" for each of the tiles. If all of the fields have the
same hard material then the cluster area for each of the fields is "21".

P1

Path <m>: 16.000

Refl. order/colour:[0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [>=10] Time <ms>: 47
Odeon©1985-2021 Licensed to: Odeon A/S Restricted version - research and teaching only! Dead balls: 0

Odeon©1985-2021 Licensed to: Odeon A/S Restricted version - research and teaching only!

P1 P1

Path <m>: 14.500

Refl. order/colour:[0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [>=10] Time <ms>: 42
Odeon©1985-2021 Licensed to: Odeon A/S Restricted version - research and teaching only! Dead balls: 0

Path <m>: 14.800

Refl. order/colour:[0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [>=10] Time <ms>: 43
Odeon©1985-2021 Licensed to: Odeon A/S Restricted version - research and teaching only! Dead balls: 0

Left: When all the individual sub-surfaces in the plane highlighted in aqua colour have same material then ODEON 17
(and later) will automatically use the sum of their area for calculation of reflection-based scattering and angle dependent
absorption. As can be seen, the large continuous area causes an almost perfect specular reflection.

Right: When different materials are assigned to each sub-surface in the plane, then these smaller areas are used for
calculation of scattering, which then becomes higher not only because of the size, but also because of the material
discontinuity at each common edge.

Estimated impulse response length and Ray-loss allowed

The Quick Estimation can now be accessed directly from the Room Setup. This makes it easy to get a
suggestion for the Impulse response length to use in impulse response calculations. The impulse

12
response length needs to be long enough in order to simulate parameters such as T 30. The value
estimated by Quick estimate will work fine for most rooms except extremely none diffuse cases such
as outdoor simulations. The value estimated will automatically be inserted in the Impulse response
length field.

is a new parameter that can also be specified in the Room Setup - see figure above. It
Ray loss allowed
determines the maximum percentage of lost rays during a simulation – due to holes and leaks in
the model. This parameter used to be 20% and was hardcoded into ODEON. The default value in
ODEON 17 is 10%, as it has become easier to make "watertight" geometries with plane surfaces
using modern modelling tools. Moreover, it is preferable to receive a warning if there are problems
with holes or warped surfaces in the geometry, which could cause leaks of rays, hence leading to
unreliable calculations.

Easier outdoor simulations

Due to major changes in the way of handling rays, in ODEON 17 outdoor simulations - like a square
surrounded by buildings in the city - have become easier than ever. If the Ray loss allowed is set to
100%, all rays are allowed to disappear out of the geometry at some point. With this option, it is no
longer needed to construct a 100% absorbing bounding box around the room geometry to prevent
termination of calculation, due to excess loss of rays.

Facilities for transferring multi-source auralisations

Some of the existing tools in ODEON have been enhanced to facilitate the handling of many sources
and receivers, from one room to another. This is typically used in multi-source auralisations, like a
symphony orchestra in a concert hall - though it can also be used for various soundscapes. More
analytically:

Source-receiver script handling

The Source receiver scripts, which allow exporting and importing sources and receiver from/to the
source receiver list, now allow point sources and receivers to be outside room limits. This makes it
possible to import e.g. an orchestra from another room, then alter its coordinates using the source
group editor (previously known as source macro and shown below), in order to place the orchestra
somewhere else.

Source-group editor
The Source group editor (previously known as source macro) has been improved to make it easy to
relocate a group of point sources. This is for example very useful when importing a complete
orchestra in order to be able to relocate that orchestra (the point sources). Functionality for
coordinate manipulation includes: rotating around a reference point, moving from one reference
point to another, move positions relative to current positions, align point sources relative to surfaces
- e.g. locate them 1.2 metres above the floor (ODEON uses ray tracing to achieve that and alignment
can be handled whether the initial source position is above or below floor) - plus some additional
options.

13
Link signal files to sources (A)(C)
The Multi-source signal auralisation expert has been equipped with two extra buttons:
• Link signal files to sources (by assigning wave file name to sources description) .
• Pair sources with signal files .

These tools help to link wave files with sources in large projects when importing into another room.
For example, a group of sources representing a complete orchestra.
The first step is to assign the sources a description equal to the name of the wave file used for that
source in a multi-source auralisation. This information can be exported and then imported through
the source receiver script into another room. The relevant wave files can be imported using the
Select multiple signal files button . Then, they can be re-paired with the sources using the form within
the Multi source signal auralisation expert .

14
Loudspeaker modelling
ODEON 17 can display loudspeaker cabinets from CLF and
EASE 3+4 loudspeaker directivity files, provided they
contain drawings. If surface drawings are included, then
cabinets will be rendered as surfaces with edges in the
3DOpenGL. Otherwise, they will be rendered simply as
wireframes. Loudspeaker cabinets will be rendered as
wireframes in all other geometry displays of ODEON. The
visualization of cabinets does not interfere with the ray-
tracing calculation, meaning that no rays will ‘hit’ its
surfaces. However, it is useful for verifying that
loudspeakers will fit at the specified locations in geometry
and that they have proper distance from the walls.

The 3DDirect Sound has now been updated to display Arrival time (delay) in ms. The delay is visualized
using reversed colour scale (upside down) which makes SPL and Arrival time maps follow the same
trend when mowing close to/away from sources. SPL and Arrival time values of the receiver below the
mouse cursor are updated in real time.

The Early arrival parameter is defined in Room Acoustic Parameter list with an inverse geography scale
- this parameters include the early reflections up to the transition order as well (and displays the
map using interpolation - if the smooth colours option is enabled for that parameter) (A)(C).

Significant improvements in Source-receiver editing

• In automated source positioning tools, such as Height experts/source macro/Tabulate source
receiver: if source(receiver) is outside the room domain, then it will be lifted inside before
dropping.
• You can turn receivers on/off in Multi-point response display by clicking with LMB (Left Mouse
Button) in the 3D View.
• You can turn sources on for the selected job in Joblist, by clicking with LMB on sources in the 3D
view.
• Direct edit of source/receiver positions in the 3D View is no longer limited to the discrete nodes
on the light blue support grid (Ctrl), but to any position in-between using the Shift+Ctrl shortcut
in the 3D Edit Source-Receiver display.

15
Customized user-interface colours
ODEON 17 comes with an option to customize the colours of the various items in the interface.
This is useful if you are partly colour blind or if you prefer using different colours e.g. for the
appearance of your reports. The User interface colours tab sheet is available from the Options>Program
Setup menu. It has the following main features:
• Any colour can be chosen for the listed functions.
• The colour combinations can be saved and loaded on different projects.
• The default colours can be easily restored using the Factory default button.

Multi-point response target curves

Target curves have been added to the measured versus simulated curves in the Multi-point response
display which consequently has changed title to Simulated versus Measured and Targets. The targets is
given in a separate file/format allowing e.g. to specify the desired minimum and maximum values
for T30 per octave for a project. The graph can be used for documenting to which degree the
simulated results fit the desired target values.

Genetic Material Optimizer

The target values must be set carefully for a genetic material optimization, as many times they
might not complement each other in a physically possible way. For instance, a high T 30 target value
of 5 sec will not work together with a high C80 value of 4 dB, simply because in real life as

16
reverberation time increases, clarity decreases. Therefore, such a combination of targets in this case
is physically impossible.

SU2Odeon version 3
The 3rd version of the SU2Odeon plugin is able to export ODEON .par geometry files to SketchUp
(SU). This allows geometries that are imported from another source (AutoCAD, Rhino, Revit) or
indeed written directly in the ODEON .par format to be corrected, altered and extended in SU and
then reimported into ODEON. This is done largely without losing information on surface data,
surface sources, grid surfaces and reflector surfaces, because surface numbers are preserved. In
some cases, SU will though create extra surfaces; an example of this is when a floor of two rooms
is divided by a wall that stands on that floor surface. Then SU will cut that floor into two surfaces.
When reading such a geometry, ODEON will manage to apply original material to all parts of the
surfaces the have been cut into multiple segments.

The plugin only supports a subset of the full .par format and all surfaces must be planar, as SU does
not at all allow warped surfaces. To make it possible to import all geometries from ODEON into SU,
the functionality File > Make .par file (and model) compatible with SU has been built in ODEON 17. This
function will convert any existing .par (and .sur) file and its associated files (.pcs, .pcg, .pcr, .pcl).
Surfaces will be renumbered, so they have the numbers 1,2,3... and only the keywords Pt, LSurf and
Layer will be used in the converted file. Warped surfaces will be modified. In cases where the point
in a warped surface is only shared by two other surfaces - then that point is relocated to agree with
the plane equations of the three surfaces. In cases where a point is shared by more than three
surfaces, extra points are generated. In any case, this process will not result in geometry that leaks
more rays - though there will be some extra edges generated which will not join each other.

Enhanced Calculation Speed

The calculation speed in ODEON 17 has been improved in two ways:
• Speed improvements in point response calculations due to enhanced ray tracing algorithms 0-
20%.
• Directivity cache makes calculations with many array speakers up to 40% faster (added to the
above number).

Improved 3D OpenGL view handling

The view in the 3D OpenGL automatically switches to camera mode (rotating/moving the camera)
when you are inside the model and to the model view (rotating/moving the room geometry) when
outside the model. In any case the shortcuts Shift+(optional Ctrl+) Left, Right, Up, Down keys are used.
The model manipulation mode has been enhanced and works similar to the 3DView (wireframe)
displays. When exiting the model, it is ensured that the model is always visible - by applying a
short rotation animation as needed.

Improved Tabulate selected source or receiver

The keep relative source/receiver height option allows positioning sources/receivers at a given
height above floor. A two-step ray tracing algorithm ensures that the tabulated sources/receivers
will be inside the room (1st step) before the new height is calculated (2nd step). Thus, the cryptic
additional drop height needs only to be specified if the room geometry contains multiple possible
floor levels.

17
Zoomable scale bar in 3D Grid (I)(A)(C)
The scale at the right side of the 3DGrid can be zoomed, un-zoomed and scrolled the same way as
other graphs in ODEON. This makes it simple to investigate a sub-range of the range auto scaled
by ODEON or defined manually in the Room Acoustics parameter list. This is typically useful if part of
a geometry has different acoustics from other parts (e.g. the reverberation time is much lower in
the class room than in the corridor. In the left image above the scale is zoomed at 1.20 to 2.10. In
the right image the zoomed grid is displayed after the mouse was released.

New audio effects (A)(C)

The AudioFX tool, available at (Tools>Audio Fxs) helps creating variations of anechoic files for
sophisticated and multi-source auralisations. The existing effects have been improved and an extra
effect has been added, called Amplitude Modulation. This effect can add more level variation
between individuals in simulations of multiple speaking people in a crowd or multiple violinists
in a string group. It works by changing the amplitude of the signal by a random envelope that is
defined between a minimum and a maximum limit in dB.

Improvements in Measurement facilities

Several improvements have been added to the measurement facilities in ODEON 17:
• Measured impulse responses, room acoustic parameters and auralisation can be normalized to target
conditions in terms of temperature (T) and relative humidity (RH%). When comparing measured and
simulated room acoustics parameters in large rooms, it is very important that this is done using
the same air conditions. For large rooms, parameters like T30 may differ considerably even at
mid-frequencies depending on RH (and less on T).
• Ideally input and output devices (eg. microphones and loudspeakers) used for measurements
should have a flat spectrum (frequency response), but this is not always the case. For that
reason, ODEON 17 is equipped with tools for creating Inverse filters to be applied in order to
compensate for the non-flat spectrum of the device.
• By default, G, Gearly, Glate and SPL are automatically hidden in the measurement results, if no
calibration or no valid calibration file is selected. The setting can be changed in the
Options>Program setup>Measurement Setup by checking/unchecking the Show uncalibrated parameters box.

18
OdeonZipArchives for auralisations (A)(C)
The menu Help>Download OdeonZipArchives with anechoic files links to the Odeon Zip Archives repository
on the Odeon home page. From this page an extensive collection of multi-source auralisations can
be downloaded. The archives include rooms, directivity files and anechoic files needed in order to
perform a multi-source auralisation. When unzipping the archives ODEON will automatically
copy directivity and wave files to the correct locations indicated by your ODEON installation. If
extra wave files need to be recreated by the AudioFX tool (see in Chapter 7) - this process is initiated
automatically too. With the tools for source handling listed above it is fairly painless to transfer a
complete orchestra defined in an .OdeonZipArchive (sources, directivity files and wave files) from one
room to another (e.g. into your own room model) a task that may otherwise be overwhelming. A
Video explaining how this is done will be available shortly on our Video Tutorials page.

Source power estimation (I)(C)

ODEON can estimate source powers for the active sources in a given job - if provided with
measured SPL in a number of receiver positions. This makes it possible to make more realistic
simulations of industrial environments where information on source power is not available from
the manufacturer (typically a problem for large machinery which cannot be analysed in a
laboratory). The tool gives adequate feedback whether a source's power seems well estimated or if
there is a need for measurements in additional receiver points or in receiver points closer to that
source.

1.4 Features of previous versions

Below we have gathered the most important information of improvements in the previous ODEON
versions. Please read more online at https://odeon.dk/product/whats-new/previous-versions.

Key features introduced with version 16

Several improvements in the interface of Reflectograms and 3D Reflection paths were made in the
single point response results. Among others, reflections can now be coloured with four different
colour scales, chosen by the user. Intensity hedgehog graphs were added in the Decay Curve tab sheet,
which allows tracing intensity components in the 3DView. A 3D matrix tool was implemented, that
calculates and displays room acoustics parameters for all combinations between source positions
(basically treating one source as the actual source and the rest as the receivers, each time).
The measuring facilities were enriched with support for ambisonic measurements, improved
auralisation of measurements, including a virtual dummy head, as well as a novel sweep correction
algorithm that compensates for a potentially poor frequency response of the output device.
In terms of modelling, a new CAD importing interface allows ODEON to read a greater range of
formats, such as .dxf, .3ds, .stl, .obj .step, .iges, .ply. files. In terms of source definition, ODEON 16 was
able to recognize EASE files directly in the point source editor.
Finally, a material archive functionality helps the user to save the 10 most recent versions of the
material list, so that any of them can be retrieved, without having to save a new version of the
whole room project each time.

19
Key features introduced with version 15
In ODEON 15 an improved 3D Grid interface made it possible to define vertical grids, individual
receiver heights for different surfaces and groups of grid surfaces (e.g. floor, 1. balcony, 2. balcony
etc.). Parallel processing was applied in more places allowing faster multi source calculations, as
well as multi source auralisation. STIPA and MTI parameters were added as well. In addition, a
statistical tool for measurements was introduced to display statistics between multiple-loaded
impulse response files. Regarding auralisation, a user-friendly interface for deriving inverse filters
for headphones was added. Using this, the actual frequency response of a headphone pair is
neutralized (flattened), so that colouration during auralisation becomes minimal. You can read
more about the features added in ODEON 14 at https://odeon.dk/product/whats-new/previous-
versions/odeon-14-features/.

Key features introduced with version 14

In ODEON 14 an improved 3D Grid interface made it possible to define vertical grids, individual
receiver heights for different surfaces and groups of grid surfaces (e.g. floor, 1. balcony, 2. balcony
etc.). Parallel processing was applied in more places allowing faster multi source calculations, as
well as multi source auralisation. STIPA and MTI parameters were added as well. In addition, a
statistical tool for measurements was introduced to display statistics between multiple-loaded
impulse response files. Regarding auralisation, a user-friendly interface for deriving inverse filters
for headphones was added. Using this, the actual frequency response of a headphone pair is
neutralized (flattened), so that colouration during auralisation becomes minimal. You can read
more about the features added in ODEON 14 at https://odeon.dk/product/whats-new/previous-
versions/odeon-14-features/.

Key features introduced with version 13

In ODEON 13 an improved interface was implemtented for the Materials list, the Source-receiver list, the
Job list, etc. They have all been equipped with sorting functions, better colour display and easier
editing properties, such as multi-delete, copy-paste, etc. Another important tool was the Genetic
Material Optimizer, an extension of the measuring system introduced in version 12. With this tool,
the materials in a room model can be automatically adjusted in order to achieve better agreement
between simulated parameters in the model and measured parameters in the real room. This makes
it possible to “tune” or “calibrate” a model before starting the acoustic refurbishment. You can read
more about the features added in ODEON 13 at https://odeon.dk/product/whats-new/previous-
versions/odeon-13-features/.

Key features introduced with version 12

In version 12, simulations and measurements were brought together. A room impulse response
measuring system was introduced, allowing the capture of impulse responses supported by a
loudspeaker, a microphone and a laptop PC. The tool is available in all editions (from Basics to
Combined) and offers easy-to-use simulations and measurements inside the same software
package. Moreover, ODEON 12 was equipped with full support for ISO 3382-3 – Open Plan Offices,
tools for grouping receivers in the multi-point response, two types of radiation in line and surface
sources, calculation of the IACC parameter (Inter Aural Cross Correlation) and many more
enhancements. Read more about them online on https://odeon.dk/product/whats-new/previous-
versions/odeon-12-features/.

Key features introduced with version 11

Parallel processing was introduced with ODEON 11, providing faster calculations on multicore
computers. The parallel processing is used for multi-point and grid responses, where many
receivers have to be calculated at a time. In addition, a Fibonacci Spiral distribution used in ray
tracing, allows an optimal distribution of source radiation pattern. A Room acoustic parameter list

20
interface allows the user to edit existing parameters and create their own ones. This is a great
feature for improving parameters or defining new ones, especially for research purposes. Apart
from these, an improved calculation setup, improved grid scaling, echo detection curves, dynamic
diffusion curves and many other features were introduced. You can read more online on
https://odeon.dk/product/whats-new/previous-versions/odeon-11-features/.

Key features introduced with version 10

ODEON 10 becomes Unicode (and utf-8) enabled, allowing text in complex character sets to be
saved with your projects, whether this is in the text files such as the geometry files (.par) or texts
composed in the various comment fields in ODEON (the material library is not fully Unicode
enabled yet). In particular, Unicode allows text in Asian character sets (Japanese, Chinese and
Korean) to be saved from within the ODEON application. It is also possible to save text which is a
mix of texts in different character sets. When upgrading to version 10 or later, the order of materials
in the Material list has been revised – please see Chapter 4. Read more on the features introduced
with version 10 on https://odeon.dk/product/whats-new/previous-versions/odeon-10-features/.

Key features introduced with version 9.1

Starting on version 9.1, ODEON is delivered with a new type of Dongle, which makes it easier for
the user to install new upgrades. This is more thoroughly described below in Chapter 1.5 and at
https://odeon.dk/downloads/dongle-license.
Version 9.1 has a far better auralisation option than the previous versions for presenting through 2
loudspeakers, called Super stereo. With the Super stereo you can make 2 loudspeakers have a much
more spacious sound, and with the right frequency response for this type of auralisation.

Some outdated directivity files such as TlkNorm.so8 have been replaced by TlkNorm_Natural.so8
which is smarter for auralisation use. There might be a risk of adding the overall frequency
response twice to auralisation output; once from the directivity pattern and once more from the
auralisation signal which inherently includes the same source spectrum (please see Chapter 6.1 for
further information). With the ‘_Natural’ version of the directivity files it is possible to obtain
correct equalization of auralisation output while also achieving correct prediction of SPL. Therefore
the ‘_Natural’ versions of the directivity files should be used when defining new sources. The old
versions of the files are kept in \old_so8\, a subdirectory to the \Dirfiles\ directory. If you wish to
use the old directivities in old/existing projects, then open the Source receiver list and click the
Repair broken directivity links button (shortcut Ctrl+L).

1.5 How to upgrade or update your current license

For ODEON (versions 9.1 and later) your ODEON software license is stored in a Smart Card based
hardware key, the Rockey 6 Smart dongle. As an ODEON user, there are the following update
situations. Please contact [email protected] if interested, to get assistance.

• Sub version update: Version number can be updated free of charge in-between the full
versions at https://odeon.dk. E.g. from version 14.1 to version 14.2
• Full version update: Your version can be updated to a full version number (E.g. version
15) by requesting a remote licence update. If you have a valid Maintenance and Support
agreement when a new version is released, you will automatically receive a remote license
update at no additional cost. If you do not have a valid Maintenance and Support
agreement, an upgrade can be purchased.
• Change edition: It is always possible to upgrade your license to a higher edition if agreed
with the sales department. Upgrade from ODEON Basics to Industrial, Auditorium or
Combined, or update from ODEON Industrial or Auditorium to Combined.

21
 • Remove time license: All new licenses are shipped with a time limitation. If your license
has a time limitation and the payment has been fully made, this time limitation can be
removed by requesting a remote licence update.
• Update user name: For copy protection purposes, your user name and country are
embedded in the dongle, but this can be changed. The user name can only be updated
separately from the other updates.

Your update or upgrade interests could be a combination of the above 4 points, which can be
fulfilled by generating a remote request file and following the steps described below. If you have
any questions about updates, please send an e-mail to [email protected].

Remote License Update or Upgrade is divided into four steps:

1.Generate a license request file and send it to [email protected] (see below)
2.You will receive a mail with the license file, when your update or upgrade has been
generated.
3. Download the received license file to your dongle.
4. Update your ODEON software installation as needed from the https://odeon.dk
homepage.
For more info on dongle update, please visit https://odeon.dk/downloads/dongle-license.

Generating remote request file

A remote request file has the extension .req. To generate a Remote request file follow these steps:
1. Attach your ODEON dongle to the USB port on the PC.
2. Run the ODEON program.
3. Check what your current license includes using the Tools|License information menu entry.
4. Generate your license request file using the Tools|Generate Remote Update Request file menu
entry - this will display the dialog shown below.
5. Email the generated request file (e.g. User2005_Dongle102009.req) to [email protected].

Once Odeon A/S has received your request file, an invoice will be processed and sent to you. When
the invoice has been paid, an encrypted license file will be e-mailed to you, and this file has to be
downloaded into your dongle for the license upgrade (see below).

Updating the dongle – downloading license file (.cif) to dongle

1. Extract the license update file from e-mail to your hard disk e.g. to your desktop.
2. Attach your ODEON dongle to the USB port on the PC.
3. Run the ODEON program.
4. Use the Tools|Download license update to dongle menu entry.

22
5. Select the file using the Select license update file (.cif) dialog (e.g. select
User2005_Dongle102009_ODEON Industrial_VXX__Restricted_0h.cif).

Update installation
When the update has been downloaded to your dongle, you can find and install the new version
or edition from the webpage https://odeon.dk/downloads/odeon-installations-updates/. You may
choose to uninstall the previous version before installation. We recommend that you keep a safety
copy of the installation file, e.g. if you at some point want to reinstall the downloaded version.

23
 Modelling Rooms
In order to make simulations of acoustics, ODEON needs a 3D model of the geometry in question.
This easiest way to do this is by using a 3rd party modelling software, eg. SketchUp, to draw the
geometry and then import it to ODEON. You can also change the geometry inside ODEON, but
this is mainly for fixes in the geometry, if the software it was made in originally is not available for
the ODEON user.

Creating new room models is probably the most time consuming task in room acoustical
modelling. However, good modelling practice will surely reduce the time used for modelling and
re-modelling rooms.

2.1 Pre-calculated Rooms

Before making any models to be used in ODEON it is good idea to have a look at the pre-calculated
rooms in your ODEON installation. You will see how a typical acoustic model looks quite
simplified – which in principle differs substantially from an architectural model, where all details
matter for visual purposes. In an acoustic model, only the most important details should be
modelled. In paragraph What to model in Chapter 2.2 you can read more about the basic rules you
should follow.

Tip: Find the pre-stored rooms in the Rooms folder in your ODEON installation. The usual path is
C:\Odeon15<edition>\Rooms, where <edition> stands for Basics, Industrial, Auditorium or Combined. You
can see all relevant paths used by your ODEON installation by choosing Options>Program setup in the top
menu bar.

Round Robin rooms

ODEON has been part of the Round Robins, where simulations where compared with
measurements for different acoustical prediction software and methods. In the Rooms folder of
ODEON you will find two rooms used as test objects in the 2nd and 3rd Round Robins (Bork, 2000;
Bork, 2005). You can check for yourself how the simulation matched the measurements, see how
below. The rooms to open in ODEON is Elmia RoundRobin2 detailed.par and PTB_Studio open curtains detailed
model.par. The geometry, absorption data, source and receiver positions as well as the measured
room acoustical parameters are those supplied to all participants in the Round Robins, by the PTB
in Germany.

• Open the room in question.

• Open the JobList (Shift+Ctrl+J shortcut).
• Select one of the pre-calculated Multi point response jobs (job one or two) in the JobList and open
it (Alt+M shortcut).
• Select the Measured versus simulated tab-sheet in the Multi Point Display.
• To select different parameters in the display, use “Arrow forward” or “Arrow backward”
• To select different octave bands, use “Arrow Up” or “Arrow down”
• To select another receiver, use the R-shortcut or Shift+R-shortcut.

24
 P2 P2
P1 3
2 1
1 2
4 P1
3
5

Odeon©1985-2005 Odeon©1985-2005

CAHRISMA project rooms

is a model of a Byzantine church in Istanbul.
Hagia Irene.par

Like the examples above, it also includes measured

room acoustical parameters from the actual site. The
church was modelled as part of the CAHRISMA
project (Conservation of the Acoustical Heritage by
8
the Revival and Identification of the Sinan's Mosques 9
Acoustics). Hagia Irene has an approximate volume 1 P1
of 39.000 m3 and RT of 4,3 sec at 1000 Hz. You can 10 7 2
3
find similar rooms in the CAHRISMA project 4 5
11
subfolder. These rooms need to be calculated. Open 6
any of the rooms, access the JobList and press ALT+A to
Odeon©1985-2005
run all jobs.

ERATO project rooms

Open the Jerash_Present_Empty.par ancient theatre and
observe the bounding box around the room. The
bounding box is not part of the actual geometry but
it is necessary in such a case where the room is not
watertight by itself (there is no ceiling). The box
encloses the geometry and allows ODEON to X
Z

perform calculations. All sides of this dummy box Y

O

are assigned 100% absorption in order to successfully

model the open air. You can find more models from
ancient theatres in the ERATO project subfolder.

Odeon©1985-2013 Licensed to: Odeon Restricted version - research and teaching only!

25
 Noise control rooms
The example Studstrup Power Plant.par, found in
the main Rooms folder, is a model of a turbine
hall at a power plant. Measured SPL(A) from
10 8 the actual site is included in this example.
7 S10S11S13
S14S16
S17
S20
6 S8S6 S4 S3
P19
P44 S9 S12S15 1
P19 P21
P2
22 11 5S18 4 S7 3 2 Apart from the common point sources, surface
20 12
19 S31
S32 S34
S35 S37
S38
S40
S36
S27 S25
18 S29 S26
P41
P22 P23
P42
S30 S33 13 sources that can model the radiation of sound
S39 S28 15 S2421
17 14
23 16 from surfaces have been used as well. This is
Odeon©1985-2007
24
due to the assumption that an engine consists
of a group of surfaces that radiate sound as
they vibrate.
Z

X
O
Y
The example Transmission rooms.par, which is
also found in the main Rooms folder includes
some walls with transmission properties and
it serves as a good demonstration of sound
insulation between rooms. You can read more
about transmission modelling with ODEON
in Appendix B.
Odeon©1985-2013 Licensed to: Odeon Restricted version - research and teaching only!

SPL (dB) at 4000 Hz >= 81.5

79.5
The room Oil rig.par is an example of an offshore
78.0

76.5

75.0
oil platform where loudspeakers have been
installed as part of a voice alarm system. Noisy
73.5

72.0

70.5

sources have been taken into considerations

69.0

67.5

66.0
M6,8
64.5

M6,10 M6,11
M6,9 M6,12
P14
P13
P12

P8
63.0

61.5
and they are modelled as surface sources.
P9 60.0
P10
58.5
P11 P23
M3,137
M3,141 57.0
M3,138
M5,142M3,140
M5,143
M5,144 M3,139
M5,146
M5,145 P24 55.5

This room has also been analysed in the

P15
54.0
P17
P19
P21 52.5
P22
P18 P16
51.0

application note Calculation of Speech

P20
M1,334
M1,335 49.5
M1,336
M1,338 48.0

M1,337 46.5

45.0

43.5
Transmission Index in Rooms which can be
42.0

40.5

39.0
downloaded from the application notes page
37.5

36.0

34.5 https://odeon.dk/learn/application-notes/.
33.0
<= 32.5
Odeon©1985-2014 Licensed to: Odeon Restricted version - research and teaching only!

Tip: Some of these rooms have been set up for very accurate calculations that can take longer time. If you
wish to carry out faster calculations, you may enter the Room setup and select the Engineering setting.
This will provide results approximately twice as fast without much loss in quality in results.

Public Address Systems

t16
t15
t14
t13
t12
t11
t10
t9
t8
t7
t6
t5
t4
t16
t15
t14
t13
t12
t16
t15
t14
t13
t12
t11
t10
t9
t8
t7
t6
t5
t4
t3
t2
t1
A14 2
t16
t15
t14
t1
t3
t2
A22 t16
t15
t14
t13
t12
t11
t10t32
t31
t30
t9 t27
t8
t7
t6
t5
t4
t3
t2
t1 t29
t28
t26
t25
t24
t23
t22
t21
t20
t19
t18
t17
t16
t15
t14
t13
t12
The model Copenhagen Central Station Array.par is
t16
t15
t14
t12
t13
t11
t7
t6
t5
t4
t3
t2
t1 t11
t10
t9
t8
t7
t6
t5
t4
t3
t2
t10 A13
t9
t8 t1 7 t13
t12
t11
t10
t9
t8
t7
t6
t5
t4
t3 t6
t2
t1 3
t16
t15
t14
t16
t15
t14 A19t11
t10
t9
t8
t7
t6
t5
t4
t3
t2
t1
A7
A6P3
A16 A10 t13
t12
t11
t13
t10
t9
t12
t11
t8
t7
t10
t9
t8
t7
t5
t4
t3
t2
t6
t5
t4
t1
t3
t2
t1 1
t16
t15
t14
t13 A9 available in the Auditorium and Combined
t12
t11
t10
t9
t8
t7
t6
t5
t4
t3
t2
t1
12 A15 8 A18 4
t16
t15
t14
t13
t32
t31
t30 t16
t16
t15
t14
t15
t14
t139
t13 t12
t11
t10
t9
t8
t7
t6
t5
t2
t1
t4
t3
13 t29
t28
t27
t24
t26
t25
t23
t22
t21
t20
t19 t12
t11
t8
t7
t6
t5
t4
t3
t2
t1t12
t11
t10
t9t10
t9
t8
t7
t6
t5
t4
t3
t2
t1 5 A21
t18
t17
t16
t15
t14
t13
t12
t11
t10
t9
t8
t7
t6 14 A17
A8
t5
t4
t3
A5
t2
t1 t16
t15
t14
t13
t12
t11
t10
t9
t8
t7
t6
t5
t4
t3
t2
t1
10 A23
editions of ODEON. This example
t16
t15
t14
t13
t12
t11
t10
t9
t8
t7
t6
t5
t4 6
15 t3
t2
t1
A12
t16
t15
t14
t13
t12
t11
t10
t9
t8
t7
t6
t5
t4
t3
t2
t1 t15
t16
t14
t13
t12
t11
t10
t9
t8
t7
t6
demonstrates a PA system, installed by Duran
A20 t5
t4
t3
t2
t1
A11 11
16 Audio, which utilizes 20 beam steered line
arrays. Several measured parameters are
Odeon©1985-2011 Licensed to: Odeon A/S
available including STI.

26
2.2 Guidelines on room modelling
You can choose between importing your rooms from SketchUp, from another CAD program, by
typing your rooms directly into a text file, or by using the ODEON Extrusion Modeller. Further
below there are some considerations that are common to the different cases, but first some
guidelines of general nature are given here.

Default coordinate system

To make it as easy as possible to operate ODEON, the
Z
following orientation of room geometries should be
applied (using a concert hall as the example): O
Y
X
• X-axis pointing towards the audience
• Y-axis pointing to the right as seen from the
audience
• Z-axis pointing upwards

Recommended size of a surface

An important theoretical consideration concerns the size of surfaces in a room model. The classical
laws of geometrical acoustics are such that for the purpose of calculating how much energy is
reflected; all surfaces are considered to be infinitely large in comparison to the wavelength. For
practical room models surfaces are not infinitely large and ODEON is, to some point, able to take
into account the limited size of surfaces in calculations – using the Reflection Based Scattering method
(see Chapter 7.5).

Still, surfaces should be kept reasonably large. Avoid using more surfaces than needed in order to
mimic the geometry. Modelling a lot of very small surfaces to achieve high geometrical fidelity will
not improve quality of results, but it will increase calculation time. It is difficult to put exact limits
on the size of surfaces which should be used; there will always be a need for small surfaces to fill
in awkward corners of the geometry.

Remember: A rule of thumb may be to keep surface dimensions larger than one wavelength at the mid-
frequencies. One wavelength at 1000 Hz is approximately 0.34 metres.

Often you will be in a situation where a few surfaces should be kept down to 0.20 metres or even
0.10 metres. You can just keep these surfaces if they are relatively few compared to the total number
of surfaces.

Tip: It should be possible to model most concert halls with a surface count in between 100 and 2000
surfaces.

Curved surfaces
All surfaces in ODEON must be (almost) planar; so curved surfaces have to be approximated by
dividing them into plane sections. But in the ODEON material list, there is a wall type called
fractional which will make ODEON treat a number of flat surfaces as one curved one. How finely
to subdivide the curve depends on the type of curved surface and how important the surface is.

Convex curves naturally disperse sound energy, so if the surface is in an exposed position (e.g. the
end of a balcony near the stage), it should be avoided, for example, to replace a quarter circle with
a single plane at 45°, as this might instead act like a reflector.

27
Concave curves naturally focus sound energy, and since focusing is a fault we wish to model, we
must try to arrange this so that it is preserved. However, this does not mean that a large number
of subdivisions are the solution. Using many surfaces in the model will:
• Make the model visually complex, and increase the probability of errors in the model, typically
small leaks may become a problem.
• Not comply with the image source theory used for the early reflections (point sources). See more
about calculation principles in Chapter 9.
• Increase the calculation time.

In order to calculate focusing from concave surfaces, it is extra important that the wall type of
surfaces forming a concave shape is set to fractional in the Materials List. If not, the concave surface
will scatter sound too much, taking into account the small areas of the individual surfaces forming
the concave shape - rather than the total area of the concave shape. The Reflection based scattering
method would produce too much scattering in this case.

Remember: Subdivisions about every 10° to 30° will probably be adequate to reproduce focusing trends
of concave surfaces, without excessive number of surfaces, thus walls in a cylindrical room may be
modelled with 12 to 36 surfaces. A cylindrical column which disperses energy may probably be modelled
with, say, 6 to 8 surfaces.

What to model
Some simple tricks during modelling will
save you a lot of time, improve the
calculation speed and offer better results,
without losing accuracy in the acoustical
geometry.

How to model an audience area?

Modelling each step between the rows in an
audience area is not recommended. The
audience area can be simplified a lot
without compromising the quality of the
results – in fact, using the suggested method
below is likely to produce better results:
Odeon©1985-2012 Licensed to: Odeon A/S

• Model the audience areas as 'audience boxes' with a height of approximately 0.8 metres above
the audience floor. In the figure above, the audience boxes are indicated by blue colour.
• Assign appropriate 'absorption material' e.g. ODEON material 11001. Y
Z

O
X

• Assign a high scattering coefficient of 0.7 to the surfaces of the ‘audience box’. See in Chapter
4how to assign scattering coefficient on a surface.
• Place the receivers around 0.4 metres above the modelled 'audience box’.
An obsolete alternative solution was to model the audience just as a flat absorptive surface on the
floor, mainly for simplifying the model for computational issues. The main problem with this
approach was that the absorption area of the room looked smaller, as more rays were likely to hit
a hard surface (like the isles between audience areas,) than it would be in real life. In order to
compensate for that, the same absorptive material had to be assigned to the isles too. Nowadays,
there is no need to follow this approach. By modelling audience as a box this has been proved to
be a realistic and reliable assumption, without affecting the computational cost.

28
How to model the podium on stage?
The same guideline as for the audience area works
Z
here. Rather than modelling each step of the
podium on stage, the podium can be simplified O
Y

into a few sloped surfaces. X

Should furniture such as tables, chairs and

shelves be included in a model of an office?
If a tabletop is close to a source or receiver point,
then it is likely to produce a strong specular
reflection towards the receiver, so if this is the case
then it should be included. However, the legs of a
chair or table should not be modelled in detail. Furniture such as shelves and screens in large office
environments, which subdivide the room – breaking up long reflection paths and introducing extra
absorption and scattering, should not be omitted. Furniture at more distant locations in the room,
which does not produce any strong early reflections to the receiver, can easily be simplified, or
even omitted from the model. This as long as the extra absorption and scattering produced by that
furniture is somehow included on other surfaces in the same regions of the room.

Orientation of surfaces – does tilt of a surface have any significance on room acoustics?
Small changes to the orientation of surfaces can indeed cause dramatic changes. Making dominant
surfaces slightly off-angle can cause extra scattering in the room, almost as if extra scattering had
actually been assigned to the surfaces in the room. A typical example is a box-shaped room where
a flutter-echo can be removed, by changing the angle of a surface by just a degree or two.

2.3 Files created by ODEON

When importing a geometry, ODEON creates a group of subsequent room files, calculation files
and result files. These should normally not be accessed outside ODEON.
All files share the same name - the name of the room - with a different extension added. Three main
categories of files are used:
• Room files, which contain data entered by the user. From these files both calculation and result
files may be re-calculated. Some of the most important extensions are .par, which describes the
whole geometry, .PcS that lists the sources, .PcR that lists the receivers and .Mat that includes
the materials used in the model.
• Calculation files, which hold data being re-used during calculations to speed up calculation (of
point response calculations). These files can be erased once job calculations have been carried
out, however if they are present, future calculations are sped up.
• Result files, containing all results. For example, the extension .GNN refers to the grid response
results in Job number NN.
See the full list of files in the ODEON help which can be accessed from within ODEON by pressing
F1 and searching for File management.

29
2.4 Importing Geometry into ODEON
The geometry import format
The file extension used for geometries in ODEON is .par. Most 3D modelling software will be able
to export something ODEON can translate into the .par format. Read about importing to ODEON
further down. The file containing the room model must be written as a text file (ASCII or UNI-
Code), having the file extension.par. The ‘old’ ODEON .sur file format is also allowed though not
described in this manual.

You can choose to create the geometry file by:

• Using SketchUp (recommended) with the SU2Odeon plugin described in section 2.5. Today
this is the fastest and simplest method, used by many ODEON users.
• Using a third-party CAD program (e.g. IntelliCAD, Autodesk Revit, AutoCAD,
3DStudioMax, MicroStation or Rhino) which is capable of creating 3D surface models and
exporting these as .dxf files as described in section 2.6.
• Using the ODEON Extrusion Modeller described in section 2.7.
• Typing the model data directly into a text file in the supplied ODEON Text Editor, using the
format described in Appendix G.
• Finally, you may combine the different modelling methods; import a CAD model from a
CAD program and extend or correct it using tools provided by ODEON.

Tip: The .par file contains geometry in the Parametric modelling language file format. This format can be
edited manually in the ODEON Text Editor (see Appendix G: Modelling rooms in the ODEON Editor).

30
2.5 Importing a model from SketchUp
Our current most preferred program for room modelling is SketchUp (SU). We offer the SU2Odeon
plug-in that allows you to make direct use of SU models in ODEON. SU is a 3D modelling software
which is operated very intuitively and creates geometries directly compatible with ODEON. It is
available in a free (educational) version, as well as a pro version for commercial use which can be
downloaded at: https://www.sketchup.com/download/all. SU also allows you to create renderings
of your rooms for impressive visualization. In order to learn how to use SketchUp, it is a must to
watch some of the introduction videos at: http://www.sketchup.com/learn/videos. In a day or less
you should have learned the basics. We have also created a few videos on modelling for ODEON
with SU at: https://odeon.dk/learn/video-tutorials/import-export/.

To make use of the SU models in ODEON, the SU2Odeon plug-in must be installed. It can be
downloaded from https://odeon.dk/downloads/su2odeon/. The download includes installation
instructions and last-minute instructions. Once you have installed SU and the plug-in SU2Odeon,
you can start creating models for ODEON using SU.

Save the SU model in the ODEON room folder and press the ODEON icon in SU menu bar. Then
SU2ODEON will create a .par file which is saved in the same folder and can be opened directly in
ODEON.

Warning: Note that for large models, the SU application may show the message “SketchUp is not
responding” while it is in fact still working on creating the .par file - normally it only takes a few seconds
but in some extreme cases it might take hours.

The models created in SketchUp and exported by SU2ODEON are inherently compatible with the
requirements set by ODEON:
• Plane surfaces.
• No repeated points.
• No surfaces without area.
• Multi loop surfaces are converted into single loop surfaces.
• Layer support.
• Automatic explosion of groups and components.
• Unique surface numbers.

Extra lines appearing in the ODEON model

When a surface is created in the middle of another surface (e.g. a window on a wall), SketchUp
automatically cuts this surface from the other surface, so that no overlap occurs. An extra line will
appear when the model is imported into ODEON (see figure below).

Y
O
X

An extra line by ODEON indicates a healthy

surface that has been cut to accommodate
Odeon©1985-2013 Licensed to: Odeon Restricted version - research and teaching only!
the window in the middle

31
This extra line is justified by the fact that each surface in ODEON needs to be formed as a series of
points towards a fixed direction (clockwise or counter-clockwise). As a result, when placing a
window in the middle of a wall, an extra line appears as a natural reaction by ODEON to help
continuing the series of points. The line is not part of the actual geometry and it is just an indication
of a healthy geometry. As can be seen in the next figure, surface No 4 (highlighted by red colour)
created from 11 points. The line formed by point 9 and 10 serves as a connector between the points
at the main wall circumference, and the points at the window circumference.

You can open the display below by opening the 3DView and pressing the N-key shortcut. Select
another surface in the list in the bottom of the window using the mouse or using the arrow keys
and press the N-shortcut again to view the points in that surface.

Remember: Always check the context-

sensitive menu that appears at the menu bar
between the Toolbar and the Options menus.
Please do so to view all functions available in
the window you are currently working in, and
their corresponding shortcuts. On the picture
on the right the 3DView window is active and
therefore the 3D_View menu appears in the
menu bar.
Remember: Surface properties assigned in ODEON (such as materials, surface sources, reflectors and grid
surfaces) are preserved when the model is re-modelled in SU and re-exported for ODEON.

Since ODEON 11, the SketchUp file (.skp file) is automatically copied with an ODEON project when
using the File>Copy files menu from within ODEON.

Creating a SketchUp model from ODEON .par file.

Using ODEON 17 and SketchUp plugin version 3 it is possible to import an ODEON .par file into
SketchUp and basically create a SketchUp model from ‘scratch’.
• Before any attempt to import the ODEON .par model into SketchUp, it is a good practice to
make it compatible with SketchUp by selecting File>Make .par file (and model) compatible with SketchUp.

32
• Open SketchUp Pro or any other desktop version down to SketchUp 8.
• Optionally delete the default person, as it normally consists of a large number of surfaces and
makes the export back to ODEON slower. ODEON will however try to remove this
automatically during if you re-export.
• Click File>import in SketchUp.
• Select the .par file of the ODEON model and click OK.

Delete!

The import process starts and might take a while. When it is finished, a ‘Save as’ dialog pops up.
Ideally keep the suggested name, so that any settings you have implemented so far in the ODEON
model (materials, sources, receivers etc.) are kept. On the right side, a list of Tags is displayed. These
are an exact translation to the layers in the ODEON model.

Layers in ODEON

The model can now be re-exported/imported into ODEON by pressing the ODEON plugin button
on the top-right.

Be aware that during the importing process, some extra surfaces might be created by SketchUp in
order to make the model consisting of exclusively planar surfaces. This can result in some new

33
surfaces in the ODEON model, without materials assigned to them (no ref.) and changes in the
order of layers (and their corresponding colours). In addition, you might experience changes in the
surfaces of Grid definition.

Remember: if you plan to edit an ODEON model in SketchUp at any point, it is a good idea to perform the
import from ODEON and re-export into ODEON in the very beginning of the project. Any extra surfaces will
be created in SketchUp for the first time and from now on you can proceed with assigning materials, sources
and jobs, without risking changes afterwards.

2.6 Importing CAD files

3D CAD programs to use with ODEON
If using models from AutoCAD it might be an advantage to use the .3ds file format for export from
AutoCAD and Import into ODEON - rather than the .dxf format. This procedure will ensure that all
relevant drawing data from the CAD drawing are exported in a way which can be used directly by
ODEON.
Generally, ODEON 16 supports not only .3ds and .dxf, but also stl, .obj .step (.stp), .iges (.igs) and .ply.

The support for the DXF file format (Drawing eXchange Format), allows import of CAD models
exported from modelling programs such as:

CAD package Web links Demo available

for download
IntelliCAD http://www.autodsys.com or www.intelliCAD.com x
SketchUp http://sketchup.com/ x
AutoCAD https://www.autodesk.com/products/autocad/overview
3DS Max https://www.autodesk.com/products/3ds-max/overview x
Rhinoceros www.rhino3d.com x
Microstation https://www.bentley.com/en/products/brands/microstation x

There are likely other programs around, capable of creating geometry data which can be used with
ODEON.

About CAD drawings

The CAD import engine in ODEON supports a number of CAD entities which can be exported from
these programs and imported directly by ODEON without any extra effort.
Depending on the modelling program used and indeed how it was used by the user, different
approaches may need to be taken into account to ensure that all, or most, of the drawing data is
exported to the .dxf file in a form which can be understood by ODEON.
If ODEON encounters entities in the import process which are not supported, but recognized, then
you will be notified.

The modelling programs should be true 3D modellers. Programs such as AutoCAD LT only have
limited support for 3D modelling and are not recommended. Programs such as Rhinoceros,
AutoCAD, IntelliCAD and 3DStudiomax are true 3D modelling programs and have been reported
to be suited for the purpose. Other programs may work as well, but in any case, you may have to
experiment in order to find the optimum way to export and import the geometries from the
programs.

34
Room models to be used by ODEON must be surface models defined from plane surfaces,
regardless if the models are created in a CAD program or if they are modelled in the ODEON
environment (e.g. using the ODEON .par format). Once a model has been successfully imported by
ODEON, it is important to perform a thorough check – geometries which look fine in the ‘drawing
program’ may still contain serious errors, such as repeated, misplaced or missing surfaces.

CAD entities supported by ODEON

Irrelevant drawing entities which are not supported
Many CAD drawings are in fact 2D “paper drawings” rather than 3D models. Such drawings do
not contain sufficient information to create a 3D surface model and are ignored in the import
process. Examples of drawing entities which are ignored are circles, dimensioning lines, texts, etc.
2D drawing data may coexist peacefully in a drawing containing useful 3D data – as stated, the 2D
data is simply ignored.
It is possible to convert a few 2D entities into model data useful for ODEON if handled properly
within the CAD program. See the 2½D entities paragraph.

BLOCK’s are supported by ODEON 16

AutoCAD native files (.dxf, .dwg) many times include block entities which are supported since
ODEON 16. In older ODEON versions any BLOCK in a drawing which contains relevant 3D surface
data must be exploded using the EXPLODE command before exported to the DXF file. ODEON will
notify the user if the DXF file imported did indeed contain BLOCK’s.

3D surface entities supported by ODEON

• 3DFACE.
• Poly meshes: MESH, WEDGE, PYRAMID, BOX, CONE, CYLINDER, SPHERE, DISH, DOME, TORUS, EDGESURF,
RULESURF and any other entities based on poly-meshes.
• Poly faces, the PFACE entity and any entity based on poly-faces.

2½D entities supported by ODEON: LINE, POLYLINE, CIRCLE

ODEON can import LINE, POLYLINE, ARC and CIRCLE entities, so called 2½D entities, when the elevation
height is set to a value different from zero using the ELEV command in the CAD program. Using the
ELEV command (at least, this is true in IntelliCAD and AutoCAD) makes it possible to convert parts
of a flat line drawing into a 3D drawing – typically a 2D floor plan can be converted into a set of
(vertical) walls. Use the CHANGE command in order to change elevation and height of these
entities from within the CAD program.

3DPOLY
As an option it is possible to import 3DPOLY (3D polylines) as if they were surfaces when these lines
are closed polygons. When ODEON exports surfaces containing more than four points these
surfaces are exported as 3DPOLY-lines. 3DPOLY-lines will not respond to the HIDE or the RENDER
commands when imported into e.g. AutoCAD, however it is possible to convert POLYLINE’s into REGION
entities which are visualized correctly as surfaces in some CAD programs (if the 3DPoly’s are not
plane, this may not work). In some cases it may be desirable to switch this import option off when
importing to ODEON. The DXF file may contain entities which the modeller did not intend to
include in the 3D surface model when being imported – the entities may have been modelled for
other reasons e.g. as assisting lines in the modelling phase.

POLYLINE (when the POLYLINE is closed and the elevation height is set to zero)
This entity is not really a true surface; however in some cases it is being used by some CAD
programs (including AutoCAD) in order to bypass the limitation of the maximum of four points in a

35
surface. If a geometry which was exported from ODEON to the CAD program has to be imported
into ODEON again, this option should be on in order to import all 3D data.

3DSOLID, REGION, BODY – recognized and supported by ODEON 16

These entities are ACIS solid modelling entities, which directly supported by ODEON 16, but not
older editions.
Solid modelling is probably the most powerful way of creating 3D surface models allowing the use
of commands such as UNION, SUBTRACT, INTERSECT, SLICE, INTERFERE etc.
With a few steps it might be possible to convert these entities into something that is understood by
ODEON 15 and older.

In Cadopia IntelliCAD Professional, the 3DCONVERT command will convert above mentioned entities into
entities recognized by ODEON (Poly-faces). It is recommended to perform this operation on a copy
of the CAD file rather than on the original.

In 3DStudioMax you select all the entities in the drawing using the Ctrl+A shortcut, then right-click the
mouse on the drawing and select the Convert to|Polyfaces (other options may also work). It is
recommended to perform this operation on a copy of the CAD file rather than on the original.

Since AutoCAD 2000, the conversion process involves exporting to a .3ds file and re-importing the
exported file:
• Export the geometry into a 3D Studio file using the 3DSOUT command (this does not change your
current CAD drawing)
• Import the just created 3DStudio file back into a new (clean) drawing in AutoCAD, using the 3DSIN
command. In this new drawing, the above entities has been converted to Polyface entities which
are supported directly by ODEON. At the same time, all entities contained in BLOCK’s have been
exploded, making them appear explicit, thus directly compatible with ODEON.

If ODEON reports of any of the unsupported entities when the .dxf file has been imported, this is
because some 3D data is available in the .dxf file in a format which cannot be converted by ODEON.
Consider following the steps above in order to create a .dxf file which can be converted by ODEON.
Do note that ACIS solid modelling extensions may not be available in all editions of the various
modelling programs.

Using layers in the CAD drawing

Surfaces will when imported to ODEON carry the name of the layer on which they were drawn.
Use the layer name to give the different parts of the geometry different names, e.g. draw the stage
floor surfaces on a layer named Stage floor, the sidewalls on a layer named Sidewalls etc.

If you are modelling subdivided surfaces such as Upper wall and Lower wall because you wish to be
able to assign different materials to these parts of a surface, it is advisable to model these parts on
different layers in order to avoid ODEON gluing these surfaces together (described below). If a
drawing is subdivided into layers, this also makes it easier to assign materials to the surfaces in the
Material List in ODEON because materials can be assigned to all surfaces on a layer in one operation.

Exporting a geometry from ODEON to IntelliCAD or AutoCAD

When ODEON exports surfaces containing more than 4 points each, these surfaces are exported
using the 3DPOLY entity whereas all other entities are exported using the 3DFACE entity. The 3DPOLY
will appear as 3DPOLY lines in the CAD program and does not respond to the HIDE and RENDER
commands like entities such as 3DFACE do. However, using the REGION command it is possible to
convert 3DPOLY’s into REGION’s which do respond to the HIDE and RENDER commands.

36
Before exporting from the CAD program
Remember that BLOCK’s are not supported by ODEON. BLOCK’s containing relevant 3D info must be
exploded using the EXPLODE command in the CAD program (in AutoCAD this may also be done by
exporting the file to the 3ds (3D Studio Max) format and importing it again as described in the
Section on 3DSOLIDS).

3DSOLID, REGION and BODY entities are not supported by ODEON, so try using one of the approaches
listed above in order to make the geometry compatible.

A final remark is that it is always a recommended practice to make backup copies of your CAD
files before making any conversions.

Performing the import in ODEON

The following steps are generic for the import of any CAD file.
• Select Files>Import from file (dxf, dwg, 3ds, stl, obj, ply, cad, step, iges) or simply drop the file on the Canvas
of ODEON.
• Specify the input file e.g. MyCADRoom.dxf.
• Specify the destination file e.g. MyCADRoom.par.

Once the file names have been specified, the Import XXX file dialog (where XXX corresponds to the
extension of the input file) appears, allowing miscellaneous import options to be specified. By
default, most of the parameters may be left untouched – however it is important that the correct
drawing unit is specified. If the geometry does not appear as expected, you may try other input
parameters.

Unit in input file

Unfortunately, many CAD files are unit-less. It is important that the correct unit in which the
geometry was modelled is selected in the import dialog. If the correct unit is not specified the
import process may fail because the geometry seems to be only a few millimetres or several
kilometres in size.

37
Geometric rules, glue surfaces
By nature, surfaces imported in the various CAD files are built from three or four sets of coordinates.
When the glue option is activated, ODEON will try to glue (or stitch, if you prefer) these surfaces in
order to form fewer surfaces with larger areas. Do note that some surfaces (based on poly-faces)
may not import correctly unless the glue option is activated (and the “don’t allow subtractive
PolyFace” algorithm is turned OFF).

ODEON will not combine surfaces with each other when they are situated on different layers in
the CAD drawing. If you wish that certain surfaces are not glued together, e.g. if upper and lower
part of a wall should be assigned different materials, then draw the surfaces on different layers in
the CAD program (preferable) or turn off the glue surface's option (may lead to an excessive
number of surfaces and may not work with poly-face surfaces).

Max. point margin

If points in the .dxf file are within this margin, the points will be considered equal. Allowing a
certain amount of point margin will allow the Glue function to perform better if the coordinates in
the model are not exact. However, if you have modelled both sides of a surface (e.g. outside and
inside surface of a balcony front edge) the Max point margin should be smaller than the distance
between these surfaces, otherwise the points on either side of the surface will be considered the
same, resulting in an undesired outcome.

Max. warp
ODEON will split four point surfaces into 2 three-point surfaces, if the surface's warp exceeds this
value. The glue option on the other hand will try to glue surfaces as long as this does not lead to a
surface exceeding the Max. warp.

Editing the imported geometry

It may be necessary, or at least desirable, to repair or make changes to geometry after import. The
3DView display, which displays the geometry when it has been imported, is a useful tool for this
purpose. Please see the context sensitive help available in this display from within ODEON for
further details (shortcut F1).

Exercise: Importing a dxf file and changing its origo Z

Try importing the file Elmia RoundRobin2 detailed.DXF, which is
located in the \rooms\ directory in your ODEON
O
installation. To make the operation of ODEON as smooth
as possible it is desirable to move the Origo of this Y
X
geometry. Once the geometry has been imported this
change may be made as follows:

Investigate the coordinates of the front edge of the stage:

1. Turn on the modelling options in the 3DView (shortcut M) and move the mouse in order to
investigate the corner’s coordinates.
2. If pressing the Ctrl key while Left-clicking, then the data for the closest corner is copied to the
clipboard – the data/text can be pasted into ODEON’s editor using the Ctrl+V shortcut.

Pasted corner data from the 3DView

Pt ?48? 10.500 -5.90000 24.00000 //for a left point on the stage
Pt ?47? 10.500 -5.90000 24.00000 //for a right point on the stage

38
We may want to locate Origo at the front of the stage. This can be done using the Mtranslate
statement in the geometry file in order to move the mid-point (average of the two points above) of
the stage to (0,0,0) – open the .par file, clicking the ODEON editor icon, then just after the ### sign
type:
MTranslate –(10.5+10.5)/2 -(-5.9+5.9)/2 -(24+24)/2

At the end of the file just before the ### sign, type MReset in order to make the coordinate system
neutral – this is desirable if adding new surfaces to the geometry later on.
Click the ODEON icon inside the editor in order to save the modified geometry and reload it into
ODEON.

Other coordinate manipulations to the geometry may be desirable; in particular the CoordSys
statement described in Appendix G: Modelling rooms in the ODEON Editor may be useful.

Tip: In the DXF import dialog, multiple options are available allowing some common changes to the layout
of the room coordinate system. The order of axis and/or the direction of axis can be changed. Also some
simple changes to the position of the origo can be performed.

Trouble-shooting
Common problems when importing are described below:

Problem with zoom or translation in the 3DView:

Model appears in a strange position on the screen and zoom /translation does not work as expected.
This problem is probably caused by some small (invisible and irrelevant) surface(s) located at odd
position(s) in the imported model.
Solution 1: Try importing the geometry once again with some of the entities unchecked (turned
off), it may be that some of the entities, such as 3DPOLY or the like were not intended to be surfaces
in the surface geometry.
Solution 2: Removing the unwanted surface:
• In the 3DView turn on the 'Modelling options' (M-shortcut), look out for odd positioned points.
• Move the mouse cursor to the position of the odd point - read one of these point numbers
• Click the ODEONEditor icon to open the .par file, remove the point and try to reload the room by
clicking the ODEON icon in the Editor. Now ODEON will hopefully report an error stating a
surface is referencing the point (which no longer exists)
• Remove that surface along with ALL the points it is referencing (out-comment them)
• Reload the room.

Problem with display of coordinate system: The blue coordinate system looks odd or behaves
strangely.
Solution: If the Origo is situated in a point far away from the geometry, the coordinate system may
not display properly when projection is turned on – in that case turn off the projection using the P-
shortcut. In order to fully solve the problem, the position of the origo should be altered as described
in the example above.

The geometry displayed in the 3DView appears to be too small or large after re-import:
If the geometry was initially imported using an incorrect unit then ODEON has defined its default
view list in the 3DView in order to display the initial version of the geometry correctly. To reset
the view list, use the Ctrl+DEL Shortcut from within the 3DView.

39
2.7 The ODEON Extrusion Modeller

A small modelling program; the ODEON Extrusion

Modeller is included with ODEON. The program is found in
the Windows start menu along with the ODEON program. It
can also be launched from within the ODEON editor.

The Extrusion modeller

allows to model “extruded
Z
geometries” in a graphical
Z
O
Y
environment, or in other O
X
words, to draw geometries Odeon©1985-2004 X
Y

using the mouse. An extruded surface is a flat 2D outline,

drawn at a specified drawing depth (the third coordinate) and
with an extrusion height. When assigning an extrusion height
Odeon©1985-2004 to the 2D outline, it becomes a holster outlined by the edges of
the ‘extrusion surface’, if so desired this holster can have a bottom and a top.
In the extrusion modeller, it is possible to make one drawing which contains multiple extrusion
surfaces, each described by a 2D outline (a simple drawing) and the line properties; drawing depth,
extrusion height, bottom check mark, top check mark and a name. If the extrusion is created in the
XY- plane (the ground), then an extrusion surface may form a volume with walls and optional floor
and ceiling, whereas other extrusion surfaces
define tables, chairs or screens.

For some geometry it may be more appropriate to

Z
draw the room in one of the other main planes (XZ
and YZ planes). As an example the auditorium.par O
Y
X
model in the figure to the right has been modelled
in the XZ plane (starting from a wall), using
separate extrusion surfaces for the room, the wall
with windows holes, the table and the windows.

Odeon©1985-2004
The Extrusion Modeller and file formats
The output from the extrusion modeller comes in two formats. The first is its own native format,
an .oes file, where the extrusion model is saved. This file can be edited and extended at a later point
in the extrusion modeller e.g. if wishing to change width of the auditorium above, to change some
of the points in the drawing or to add other features.

The other format is the ODEON .par format which is loaded into ODEON for calculations. The
parametric format cannot be edited in the extrusion modeller; on the other hand it may be manually
edited to any degree of freedom if needed and it is possible to make use of the benefits of the
parametric format described in Appendix G: Modelling rooms in the ODEON Editor, e.g. when
modelling geometric shapes such as cylinders and domes, when combining 2 extrusions in
different planes, or when it is appropriate to describe parts of a geometry using parameters. The
.par file can be edited in the ODEON Text Editor. The 3DView available in ODEON is a useful tool when
investigating or modifying an already existing file; load the .par file into ODEON, study the room
in the 3DView – please see the help text (shortcut F1) available from within this display, then make
the changes in the editor which can be opened from within ODEON.

40
Using the Extrusion modeler
Start the program; a shortcut to the program is found at the Windows Menu
Start|Programs|ODEON|ODEON Extrusion Modeller.

Initial settings
Before starting to model geometry, select the
drawing plane (XY, YZ or XZ) which is best
suited for the geometry to be modelled. Also
select properties for grid and snap spacing.

Using a drawing of the floor plan as the layout

It is possible to load a 2D background drawing
section plan as basis for the geometry in various
image formats (png, jpg, gif, emf, wmf,) and to
use this drawing as the basis for the drawing.

To load a background drawing

• Use the File|Load backgound drawing menu to locate your image file.
• Double click on your desired origo in the drawing as requested (could coincide with an
intersection between a horizontal and a vertical module line).
• Double click at a point having some horizontal distance from the origo as requested (could be
another vertical module line).
• Enter the distance between the two points (e.g. distance between vertical module lines).
• If the drawing is very dark, this may be disturbing, use the File|Make background drawing lighter to
lighten the drawing. (This can be repeated).

The drawing has now been scaled and is fixed to the drawing canvas - when scrolling the drawing
canvas or zooming, the drawing will adjust appropriately. If not satisfied with your scaling of the
drawing, repeat the process above.

Drawing an extrusion surface

Left click the mouse at the positions where the points in the surface are desired – if no points and
lines are generated, then you need to bring the current surface in edit mode using the Insert (or Esc)

41
shortcut or by double clicking a point in the surface. Once all points in the surface have been
defined, finish the current extrusion surface by starting a new one, using the Ctrl+A shortcut or
pressing the Insert or Esc shortcut. To assign a drawing depth (X,Y or Z) and an extrusion (dX, dY or
dZ), select the surface in the Surface editor table where it can also be specified whether the extrusion
surface should have a bottom and a top and where a Description may be entered. The drawing depth
and extrusion for each extrusion surface is displayed graphically at the bottom of the application
window.

Editing or correcting an extrusion surface

In order to make corrections to an extrusion surface, select it in the Surface editor table or click one of
its points in the drawing area, then bring it into edit mode using the Insert (or Esc) shortcut. Once in
edit mode, it is possible to change coordinates of the points, insert or delete points, and to move
the surface using the mouse operations listed below. It is also possible to enter the precise
coordinates of points in the Point editor table which list the points in the selected surface, so it is an
option to draw a sketch using the mouse and then fine tune the coordinates afterwards in the Point
editor table.

Operation on surface Mouse operation

Create a new point in selected surface LeftClick mouse

Select the point in selected surface, which is closest to Mouse pointer Ctrl+LeftClick mouse

Move closest point in selected surface Ctrl+Alt+LeftClick mouse

Move selected surface Shift+LeftClick mouse

Manipulating the viewport

The viewport can be manipulated using the shortcuts listed below. It is possible to make changes
to the view while drawing a surface.

View operation Mouse operation

Scroll drawing area Right mouse button
Zoom (In/Out) Alt + Left mouse button

Snap to grid
Snap to grid enables points (new points or points being moved) to be positioned exactly at the
intersection of the grid lines. In special cases the point can also be inserted at only one grid line
when the other coordinate is that of an existing point (see below, snap to existing points).

Snap to existing points

Snap to existing points enables points to be precisely located on either or both reference coordinates
of existing points.

The snap point

The snap point is a special case of snap to existing points. In some cases, you may want to move a
surface to a precise location e.g. (0.33, 0.46) not being a point on the grid nor an existing point of
another surface. In that case:
• Create a new surface.
• Click the approximate position of the reference point.
• Change the coordinates to the exact position, e.g. (0.33, 0.46) in the Point editor.
• Press Insert (or Esc) to finish editing the surface, the point will appear with the mark +Snap point.
• Select the fix point in the surface to be moved (left mouse button) and move it to the location of
the snap point. Do note that Snap to existing coordinates option must be checked.

42
Once the surface has been moved, the surface containing the snap point may be deleted; this is not
a strict requirement as surfaces containing only one point, will not be transferred to the .par format
to be used in ODEON.

Coordinate snapping
When entering new points (clicking the mouse), coordinates of points will be truncated to the
nearest snap, e.g. if the snap is 0.25 and the X-coordinate of the clicked position was 0.15, then the
X-coordinate of the point will be 0.25, if the Snap to grid option is enabled. But it is still possible to
fine adjust the coordinates in the Point editor (lower right corner of the Extrusion modeller). If Snap to
existing points is enabled, then the X,Y (or Z) coordinates of already defined points are also snap-able.

Ortho snap
Enable Ortho snap if the surface to be drawn has many edge angles of 90 (or 270) degrees. The option
can be turned on and off while drawing a surface.

Lock equal points

When this option is enabled it becomes possible to move points in multiple surfaces if the X,Y (or
Y,Z or X,Z) coordinates are equal. This is useful for multi-level rooms where e.g. X and Y
coordinates are identical for multiple points while the Z coordinate differs.

Snap to existing coordinates

Makes it easy to draw new points making use of horizontal or vertical coordinates in existing points
– also when these do not match current snap size.

Lock H and V snap

If this option is checked then Horizontal snap size will automatically be updated when the vertical
snap size is changed and vice versa.

Relative or absolute extrusions

Use the Ctrl+H shortcut to toggle between relative or absolute extrusions in the Surface editor table.
When extrusions are displayed in relative measures an extrusion may be defined as Z=10 and dZ=5,
telling that the extrusion starts at a height of 10 and has an extrusion height of 5. If toggling to
absolute extrusion then the same extrusion is displayed as Z1=10 and Z2= 15 telling that it starts at
Z=10 and ends at Z=15.

Modelling an array of surfaces

Each extrusion surface has a set of array properties associated with it (Nx, Distx, Ny, Disty, Nz, Distz),
a set for each of the three main orientations in the room. These properties can be found in the Surface
editor and define how many times the surface should be repeated in each of the main directions and
the distances between the repetitions. This feature is typically used in order to create a number of
columns, beams, tables or chairs with a regular spacing. When editing an array surface e.g.
modifying a point, all the repetitions of the surface will be changed accordingly.

Exploding an array of surfaces

If individual changes are needed, the arrayed surface must be ‘exploded’. Once this operation has
been carried out, the surfaces in the arrayed surface have been turned into individual surfaces
which can be modified surface by surface (e.g. delete some of them). It is not possible to perform
the reverse of the explode operation so before exploding an arrayed surface, make sure all
operations common to the surfaces in the array have been carried out.

43
Define Layers
For later management of acoustics materials inside ODEON, the extrusion modeller has layers.
This allows you to easily assign materials to all surfaces on one layer in the Material list within
ODEON. Create a layer with a name and colour in the layer box (above the surface editor).

You can deactivate layers to make them invisible in the extrusion modeller but
deactivated/invisible layers will still be loaded into ODEON.
The current layer (checked as Current in the Layer list) is the one that new surfaces are created on.
By checking Move all objects on layer it is possible to move multiple surfaces in one operation. It is
possible to change the layer of an extrusion surface in the Surface editor allowing you to group
surfaces on one layer prior to moving them.

Modelling a chair
As an example, we will create a chair with the seat dimensions 0.4 x 0.4 and a back rest with the
height of 0.4. Legs and other small details should be omitted. Therefore, if modelling a room in the
XY Modelling plane then a chair may in effect be considered as an extrusion which excludes the
top and three of its sides. To make things easier when modelling through the extrusion modeller,
do the modelling around origin, then move the chair to its final location when finished. When
modelling around the origin, it becomes easier to read the dimensions of the seat of the chair and
to use grid and snaps without the need to calculate dimensions of the seat:
• Set the grid and snap size(s) to 0.4 metres.
• Click on 4 points of the grid, starting from x = 0 and y = 0, to form a square (seat of the chair).
• Press Insert or Esc to stop drawing (Insert or Esc toggles point input on/off). The last dotted line
should change to a solid line.
• Change the Z coordinate to 0.4 metres in the Surface editor to define seat height.
• Change dZ to 0.4 metres in the Surface editor in order to define the height of the back of the chair.
• Uncheck Top in the Surface editor to disable the top surface.
• Uncheck 3 points in the Point editor to disable extruding those sides. Only one side will be
extruded (back of the chair).
After following the steps above, the extrusion modeller window should look like the first figure
below. Then, clicking the ODEON icon in the Surface editor produces the second figure below.

44
 Creating a circular surface
To create a circular surface, first draw a line (a surface with two points) in order
to specify centre and radius of the circle, then use the Ctrl+O shortcut to activate
the circle tool and accept to create a circle from 12 points. If the circle tool is
clicked when the selected surface contains less than two or more than three
points then a help text is displayed, this text will also explain about ellipses.

Make a circle into a half-circle

Delete the 5 upper points in the circle in order to reduce the circle to a half-circle.
At this point you should have created the half circle in middle of the Figure
above.

Mirroring the surface

Following from the creation of a half-circle, the mirror functionality can be used
to create a table plate with circular ends; select the horizontal mirror and specify
the coordinate of the mirror line (in the Figures the coordinate was -1.00). Select
the first of the two points to be connected across the mirror line and finally use
the Mirror shortcut Ctrl+M to create the full table. The position of the mirror is
easily changed if holding down the Shift key while pressing Right-mouse button
and moving it – if performing a very significant move in the horizontal or
vertical direction this will toggle between a horizontal and a vertical mirror line (blue dashed line).

Mirror manipulation Mouse operation

Move mirror line /toggle between vertical and horizontal Shift+Right mouse button

mirror line

Creating a mirrored copy

Uses the current mirror line in order to create a copy of the currently selected surface

Create scaled copy of surface (Alt+Ctrl+C shortcut)

As above but a dialog appears, allowing input of a scaling factor.

Create copy of surface

Will create a copy of the selected surface. The copy is offset slightly from the original one to make
it visible.

Rotating a surface
To rotate a surface, select the point in the surface in which the surface should be rotated, then
activate the Rotate dialog using the Ctrl+R shortcut and enter the number of degrees to rotate the
surface (positive rotation angles always counter-clockwise - CCW). A surface cannot be rotated
around a point which is not included in the surface. However, you may rotate the surface around
an arbitrary point with the trick explained below.
• Select the surface.
• Bring it into editing state (Esc or Insert shortcut).
• Add the rotation point (it is not important where it is inserted in the sequence of points).
• Rotate the surface.

45
• Delete the rotation point from the surface (Del/Delete shortcut).

To rotate a surface around a point which is not included in the surface, a) Insert a rotation point in the surface, b) Use
the Rotate surface shortcut Ctrl+R to activate the rotation dialog and specify the rotation angle – in this case 25 degrees,
finally delete the rotation point from the surface.

Examples
A few examples on extrusion models are installed with ODEON – the examples are located in the
\ODEON<Edition Name>\rooms\oes Directory. The best way to learn about benefits as well as limitations of
the extrusion modeller is to load the examples, investigate the surfaces (e.g. scrolling the point and
surface tables) and to load the models into ODEON. This way the models can be investigated when
being extruded.

Special extrusions
There are a few extrusion surfaces which are treated differently by the extrusion modeller:
1. A surface with an extrusion height of zero will produce one and only one ‘horizontal’ surface
no matter if a Bottom or Top surface is selected.
2. An extrusion surface which only contains two points will only produce one ‘vertical’ surface,
no matter if Bottom or Top surface is produced. If only a single edge is extruded into a surface
with a height of zero, then no surface is produced.

2.8 Model check in ODEON

When you open a room file with Files|Open Room model, ODEON will check the validity of the file.
However, this only checks whether the format is correct and the data is consistent, not if a
meaningful geometry is defined. Furthermore, it will not check if the model is watertight, or
conversely if there are leaks. So, after opening a room file, you may start by verifying if the
geometry is meaningful and without errors.

This may involve:

• Viewing the room in a 3DView.
• Viewing the room in the 3DOpenGL display.
• Analysing the geometry for unacceptable surface warps in the 3DGeometry Debugger.

Analyzing the geometry for unacceptable surface overlap in the

3DGeometry Debugger
• Checking for missing surfaces in the room (forming holes in the geometry). The Unique edge's
function available from the 3DView may help you (shortcut E).
• Testing water tightness of the room, using either 3D Investigate Rays or 3D Billiard.

Viewing the room in a 3DView

The 3DView , displaying your room once loaded into ODEON, has a large number of features which
can be useful when creating and verifying geometries for ODEON.

46
 Remember: Always check the
context-sensitive menu that appears
at the menu bar between the Toolbar
and the Options menus. Please do so to
view all functions available in the
window you are currently working in,
and their corresponding shortcuts.
On the picture on the left, the 3DView
window is active and therefore the
3D_View menu appears in the menu
bar.

Viewing corners and coordinates in a selected surface using the N

shortcut and highlighting corner and displaying the coordinates of
the corner closest to the mouse pointer using the M- shortcut.

The perspective option (shortcut P) allows you to turn off

the perspective of the room, to get an isometric display of
the room. This may prove valuable when investigating
warped surfaces.
The Unique Edges option in the 3DView display shows
edges, which only occur on one surface. Such an edge is
"free"; it might be the edge of a free-hanging reflector, but
it also could be the result of an error whereby two
surfaces, which should join along an edge, do not.

Example:
Modelling a box-shaped room consisting of 6 surfaces, but forgetting to define the 6th surface in the
geometry file. This room will have a hole where the 6th surface is missing. The unique edge's
function will show where the missing surface should have been.

Using the 3D OpenGL display for model verification

The 3D OpenGL display is very useful for detecting holes in the geometries. Especially if
stepping outside the model (Back-arrow shortcut) and rotating the model using the Ctrl+Arrow
shortcut. See the corresponding 3DOpenGL dropdown menu for more shortcuts. When materials
have not been assigned to all surfaces in the room, surfaces will appear in random colours making
holes easier to spot. If materials have been assigned, the colours will by default reflect the acoustic
properties of the surfaces, however it is possible to turn on the random colouring at will, using the
R-shortcut. If the model contains layers it is also possible to show the layer colours using the Ctrl+L
shortcut.

Remember: The 3DOpenGL may occasionally fail to display complicated surfaces, including numerous
holes correctly (typically surfaces created by CAD software using solid modelling techniques and
subtractions) although the surfaces are perfectly legal with respect to ODEON. In these rare cases you
may assure yourself that the model is in fact correct by putting point sources at various test positions
and by conducting tests using the 3D Investigate Rays and 3D Billiard utilities.

3DGeometry debugger
Overlapping and warped surfaces should be avoided in the room model specified in the geometry
file, but a certain amount of overlap and warp (by default 50 mm) is allowed without generating a
warning. By overlapping surfaces is meant surfaces, which define a part of the same plane in space.
In the simple case this can be because the surfaces are simply duplicates, another case could be a

47
door, which has been defined in the same plane as the wall in which it is mounted. Overlapping
surfaces should be avoided because it will not be clear which absorption coefficient should be
applied at a reflection in case of overlapping surfaces with different materials.
Warps can lead to "holes" in rooms at edges of joining surfaces, with erroneous results as a
consequence and the surfaces will not be well defined.

Using the 3DGeometry debugger in ODEON, a list of warnings and a corresponding illustration
in a 3D display will be generated, whenever an overlap or a warp exceeds the value specified in
the Room setup|Model>Air conditions dialog.

Overlapping surfaces are a tricky problem because it is usually invisible on 3D projections of the
geometries, however such errors in the model may lead to unpredictable results, so always check
models of considerable complexity for overlapping surfaces.

Testing water tightness using 3D Investigate Rays

Testing a new model for "water tightness" (i.e. whether it is completely closed) may be done using
a 3D Investigate Rays window .

The room model may not be watertight if:

• Surfaces are missing from the model.
• Surfaces are unacceptable warped.
• Boundary surfaces have been assigned transparency coefficients greater than zero.
• Boundary surfaces have been assigned Material 0 transparent.
• Sources are located outside the room.

Before investigating ray tracing, you will have to:

• Make the boundary surfaces of the room "solid" by assigning materials to them. For the moment
it does not matter what the materials are, as long as they are not transparent ( Material 0) or fully
absorptive (Material 1). Go into the Materials List and assign, e.g. 20% absorption to all surfaces (use
Ctrl+Ins to do this in one keystroke).
• Place a source somewhere inside the room. Sources are defined from the Source-receiver List. At
first it may be a good idea to define a point source somewhere in the middle of the room.

Open a 3D Investigate Rays display and run it with, e.g. 1000 rays, with a Max reflection order of zero. This
will test whether any holes can be seen from the source-position and should reveal any large
problems. The tracks of lost rays will show outside the room boundaries, and indicate whereabouts
in the room, the problems occur.

If rays are being lost, and you have an idea of which part(s) of the room is /are "leaky", a number
of things may be done:

• Reduce the value of Max. accept. warp in the Room setup at the Model|Air conditions page. Then run the
3DGeomtry Debugger. Warnings will appear if surfaces have a warp or an overlap above the
acceptable range. This may reveal slight warps of surfaces in the leaky region of the room which
then have to be reduced as far as possible by revisions to the geometry file.
• Use the 3DView or 3DOpenGL for inspection of the model to study the region(s) under suspicion. It
may turn out that a surface is missing or does not join to its neighbours in the expected manner.
It may help to zoom regions in question with the Highlight surfaces, Show corner numbers and coords and
Modelling options switched on.

48
2.9 Combining geometries
It is possible to combine geometries imported from an external CAD program with geometry
modelled in the Extrusion modeller or modelled in the parametric modelling format of ODEON.
When a geometry has been imported in ODEON it will have the .par format. This is a text format
and the content of different room-files can be copy-pasted together to one room file. This is done
in the Odeon Editor. Just copy the content in the Odeon Editor in one room, add in the Odeon
Editor of another room and press “Save and Reload in Odeon” (the ODEON symbol in the Editor).

Be aware that you need each surface number to unique. This, and other issues, can be solved by
some facilities in the parametric modelling format: NumbOffset, CoordSys, Unit, MTranslate, MRotateX,
MRotateY, MRotateZ, MScale, MReset and MPop.

Below is an example/outline which illustrates how a number of geometries can be merged together
into one parametric file:

###
CoordSys X Y Z
.....1. model data.......
.....1. model data.......

NumbOffset 1000 ;avoid reusing point and surface numbers which have already been used
CoordSys Y X Z ;swap coordinate axes if needed
MTranslate 0 15 20 ;Translate /move geometries as needed
.....2. model data.......
.....2. model data.......

NumbOffSet 2000
MReset ; restoring default origo
CoordSys X Y Z; restoring default coordinate system

..…More model data ???

###

49
 Sources and Receivers
Once the geometry of a room has been loaded into ODEON, sources and receivers have to be placed
through the Source receiver list (shortcut Shift+Ctrl+S), as described in the Quick Start Guide
(available online at https://odeon.dk/learn/quick-start-guides/). This manual will not repeat the
content of the Quick Start Guide, so please read that first.

The sources could be an omnidirectional point source, a person speaking, a loudspeaker, a road
with traffic, a large industrial machine, or something else.
Depending on the sound source you need, four different types of sources are available in ODEON:

• Point Sources. Available in all editions. Most common type of source where sound radiates
spherically from a point. It is possible to define the directivity pattern of a point source, so many
cases can be simplified to the use of point sources. Used e.g. for omnidirectional point sources,
for people speaking, for instruments or simple loudspeaker setups.

• Line Sources. Available in ODEON Industrial and Combined. Sound is radiated from a line. Used
e.g for roads with traffic or pipes with running water.

• Surface sources. Available in ODEON Industrial and Combined. Sound is radiated from one or
multiple chosen whole surfaces in the room. Used for e.g. industrial machines or crowds of
people in restaurants.

• Array Sources. Available in ODEON Auditorium and Combined. Used for loudspeaker array
setups.

The receiver has a very simple form being just a “point” mono-microphone. For auralisation (see
Chapter 7) the exact direction of the head is needed, this is something which can be adjusted in the
JobList.

3.1 Generic point sources

Generic point sources such as the OMNI or SEMI directional sources are typically used for calculating
the frequency response and the parameters characterizing the room-acoustics. It is typical for
generic sources which are defined mathematically.

3.2 Natural point sources

With “natural sources”, we refer to sources such as human voice, an acoustical instrument or
similar. Natural sources are typically used for auralisations and/or calculation of acoustical
parameters that depends on a specific sound power of a natural source. E.g. speech intelligibility
of a person or sound pressure level by a smaller machine.

50
A recorded signal for auralisation is
associated with the directivity pattern of the
actual source presented during the recording.
In an auralisation, the directivity pattern for
natural sources in ODEON is used together
with the recorded frequency spectrum of e.g.
a voice. If this isn’t handled correctly, this will Signal from anechoic
result in an auralisation where the overall Directivity with spectrum recording with
of natural source spectrum of source
frequency response is included not once but
twice; first time through the directivity
pattern which includes the overall frequency
response, second time through the recorded Calculating reflections

source signal, which inherently includes that

response. Calculating
Adding acoustical
For example, in order to estimate SPL or STI parameters
inverse
correctly and for calculation of acoustical spectrum
parameters, it is desirable that the true
frequency content is included in the
directivity pattern. However, for an
Calculating BRIR
auralisation, the directivity pattern should be
equalised with the inverse spectrum of that
recorded at the font axis of the natural source signal (i.e. the wave file with human voice recorded
with a microphone at the front axis).

From ODEON version 8.5 and up, it is possible to create correct estimates of parameters from
natural sources - and at the same time create correct auralisation, where the overall spectrum is
included only once. It is necessary to use a source marked Natural. ODEON has some directivity
patterns which have the word Natural attached to their names e.g. BB93_Normal_Natural.So8. When
natural directivity patterns are selected from within Point Source Editor, a green natural-label is
displayed next to the equalization entry fields.

If having existing directivity patterns of natural sources which are not marked natural, this can be
done by using the Tool>Directivity patterns|Mark So8 file as natural directivity. When creating new
directivity patterns, this information is part of the input data.

Always use the _Natural versions of the directivity files when defining new natural point sources.
The old versions of the files are kept in \Old_So8\ a subdirectory to the \Dirfiles\ directory. If you wish
to use the old directivities in old /existing projects, then open the Source receiver list and click the
Repair broken directivity links button (shortcut Ctrl+L).

Samples on natural directivity patterns (BB93_Normal_Natural,

BB93_Normal_Raised and Soprano ref. 42).
The BB93_Normal_Natural source type (BB93, 2015) corresponds to a male speaker with a normal
vocal effort. The gain and EQ fields in the Point source editor (inside ODEON) should be set to zero. This
source is also a reasonable approximation to a female speaker, except that the 63 and 125 Hz band
should be ignored.

To simulate a trained speaker addressing an audience in a raised voice, use the TLKRAISE source.
This has the same directivity as TLKNORM, but the levels in the eight octave bands are respectively 2,
2, 5, 7, 9, 8, 6 and 6 dB higher.

51
The directivity pattern of Soprano ref. 42 is the directivity of a soprano singing opera (Parati, et al.,
2003).

3.3 Common loudspeaker format, CF1 and CF2 files

ODEON version 8 and up supports the Common Loudspeaker Format
which is an open format for loudspeaker data, supported by several
loudspeaker manufacturers as well as manufactures of software
programs such as ODEON. The Common Loudspeaker Format was
developed and also maintained by the CLF-group at
www.clfgroup.org. It is an open, though secure, file format for
loudspeaker performance data and polar plots. Loudspeaker manufacturers can use the CLF
format to supply data to end users of professional room acoustic computer programs.

CLF files are binary distribution files having the extensions CF1 and CF2. These file are created by
loudspeaker manufactures or independent measurement services which have obtained a license
key for the CLF authoring kit (http://clfgroup.org/author.htm). The need for an authoring key for
the authoring kit ensures that data distributed in the CLF format can be traced back to the author
and that data can’t be modified by third party. In order to view all data in the CLF format you can
download a free viewer from the CLF group home page.

CF1 has a frequency resolution of 10 degrees, 1/1 octave and CF2 has a frequency resolution of 5
degrees, 1/3 octave. If data are available in either format for a selected loudspeaker then the CF2
format should be preferred because of its angular resolution. Currently ODEON does not make
use of the higher frequency resolution of the CF2 format except for loudspeaker array modelling
where ODEON makes use of the higher frequency resolution as well as of phase data (if available)
for calculation of distance dependent directivity of the loudspeaker arrays (which are composed
from multiple units added with phase).

The CLF Group is providing a set of free tools for data editing, conversion from text to binary
format, and viewing binary data. This allows loudspeaker manufacturers and independent
measurement services to create, view and verify binary distribution files and ensures that it is easy
for loudspeaker manufactures to make data available. Links to loudspeaker manufacturers
currently providing binary distribution files can be found at the download page at
www.clfgroup.org. If apparently the data of interest is not available from a manufacturer of
interest, then assist the CLF-group by encouraging the manufacturer to make such data available
– free tools for this purpose can also be obtained at the CLF group’s homepage see
http://clfgroup.org/clfdocuments.htm and http://clfgroup.org/author.htm.

3.4 Directivity files in the EASE 3 and EASE 4 formats

Since ODEON 16, .Lob + .Spk directivity files (EASE 3 and EASE 4) are supported. These files contain
directivity patterns in 5 degrees resolution, 1/3 octave. These data are supported for use for simple
point sources where they provide and angular resolution of 5 degrees and a frequency resolution
of 1/1 octave. Currently ODEON does not support the use of .lob and .spk files for loudspeaker array
modelling. A set of directivity in the EASE 3 and 4 formats may come with a number of files having
the extensions .lob, .spk, .fed, .ffc, .fvt and .phs. Currently ODEON only makes use of .lob and .spk files
which defines the directivity balloons and the sensitivity. Both the .lob and .spk files must be present
for ODEON to recognise and use the given directivity. It is though recommended that all the files
are copied to the ODEON’s DirFiles folder, as they contain extra data, like phase information, geometry
of the speaker etc. ODEON 17 is able to display a drawing of the geometry directly in the Source
3DView.

52
Unlike the CLF data, the .lob and .Spk data formats does not ensure data integrity, it is not possible
to see if data was created by loudspeaker manufactures or independent measurement services or
if it was created by someone needing the data and entering some data themself – therefore it is
recommended that these data format should only be downloaded from trusted source (e.g. directly
from the homepage of the loudspeaker manufacturer. If data is available in the CLF format then
these data should be preferred! – and again you are urged to prompt the loudspeaker to make data
available in the CLF format if they are not.

File location for directivity files

Regardless if files are in the .cf1, .cf2, .lob + .spk or in ODEON’s native formats (.So8, SoCF1Tab or
SoCF2Tab), the files should be stored in ODEON’s directivity directory as specified inside ODEON
at Options>Program setup>DirFiles folder. The files may be stored in subdirectories to this directory
allowing loudspeaker directivities of different brands to be located in separate directories e.g.
C:\ODEON...\DirFile\ManufacturerA or C:\ODEON...\DirFile\ManufacturerB. We have taken the opportunity to
create a number of folders for manufactures who supply loudspeaker directivity files in the CLF
format. Using these predefined directories, it is easier to move a room from one PC to another
without breaking file linkage.

3.5 Creating directivity patterns in ODEON’s .So8 format

Tools for creating directivity patterns in the ODEON .So8 format can be found at the Tools>Directivity
Patterns>Create directivity pattern in plot editor menu entry inside the ODEON program. The tools allow
you to expand the set of source directivity pattern files available for point sources in ODEON. The
ODEON directivity pattern file (Version 3 or later) contains information on the sound levels for the
eight frequency bands 63 Hz to 8 kHz in dB for each 10° azimuth and 10° elevation. These files are
binary and have the extension .SO8. An example on a directivity pattern is the pattern stored in
OMNI.SO8.

Entering a directivity plot using the Directivity polar plot editor

The easiest way to enter a new directivity plot is to use the built-in polar plot editor which allows
building a directivity plot from a vertical and a horizontal polar plot. Click Tools>Directivity
Patterns>Create directivity pattern in plot editor. The following window appears:

53
Enter the dB values (sound pressure level) for the horizontal and vertical plots at the selected
frequency band in the corresponding tables. The angular resolution is 100 degrees. If data is not
entered for all angles (e.g. if the data are not available) ODEON will do interpolation between the
angles entered. For angles between both polar plots, ODEON will perform elliptical interpolation.

Equalization
If the source to be used in ODEON has a fixed frequency dependent sound power level, the
equalization option in the Polar plot editor is used for entering the dB values.

Calibration
Three different options are available in the Polar plot editor:
• If No calibration is selected, ODEON will use the SPL dB values as entered in the polar plot table,
adding the equalization values entered. This option is typically used if the source is a
loudspeaker.
• If Calibrated source is selected, ODEON will add the equalization if entered, then shift the resulting
SPL's of the source in order to obtain a reference sound power level of 0 dB re 10-12 Watt at the
selected Calibration frequency band (typically 1kHz). This calibration type is typically used when a
generic source with an adjustable level is needed for calculations, such as the OMNI or SEMI
directivity pattern,
• If Sensitivity calibration is selected, the SPL's of the source will be shifted in order to obtain the SPL
in dB on front axis of the source at the distance specified as Ref. distance metres at the selected
Calibration frequency band. This calibration type is typically used if the free field sound pressure
level measurement is available for a natural source.

Maximum dynamic range

The minimum level will be Max level minus Maximum dynamic range. If the range is large the display
may be difficult to interpret /view in the directivity viewer. The source will usually have its max
levels on its axis.

Odeon directivity text import format

Normally the text format should not be necessary to use, as most common sources are defined
already in ODEON or available on www.clfgroup.org.

The data entered in the text format should be in relative calibration across frequencies, but
doesn’t need to be in any absolute calibration. The absolute calibration is an option, from within
the program as described under the section below: “Creating a new directivity pattern using a
text file as input”.

The first non-comment line of the input file indicates whether the data is for:
• POLAR set containing horizontal and vertical polar plots, for sources where only a horizontal and
a vertical plot are known, e.g. a loudspeaker.
• FULL set, for complex sources where directivity data is known for each 10° Azimuth and 10°
Elevation.
• SYMMETRIC set, for symmetric sources, e.g. a trumpet.

The second non-comment line in the input file indicates whether the data represents a Natural
source such as a musical instrument, or a person speaking - and also if it represents a source such
as a loudspeaker which does not include the frequency response of the source signal. The purpose
of this Boolean is to make auralisation correct for natural sources set NATURAL to:
• TRUE for sources such as musical instruments.
• FALSE for sources such as loudspeakers.

54
Each of the subsequent lines of the input file should contain sound pressure levels in dB for a
complete 180° of elevation (from the forward axis to the backward axis). The resolution must be
10°, hence each line contains 19 values (0°, 10°, 20°, …, 160°, 170° and 180°).

Lines containing comments, and empty lines, may be inserted anywhere in the ASCII input file, as
long as they do not come between data items, which should occur on one line. Comment lines must
begin with a colon (:), a semicolon (;) or an asterisk (*).

When only horizontal and vertical polar plots are known (POLAR)
The first non-comment line of the file should start with the word POLAR. In the polar case, there
are four lines of data for each frequency band. The first four lines are for 63 Hz, the next four for
125 Hz, and so on.

For a given frequency, the first and last values must agree on all four lines, since all the polar plots
meet at the front and back of the plot. The first line of a group of four is the upward vertical polar
plot as seen from in front of the source (12 o'clock plot). Then the left horizontal plot (9 o'clock plot)
appears downward vertical plot (6 o'clock plot) and finally the right horizontal plot (3 o'clock plot).
As a minimum there must be 1 + 4 * 8 lines in a polar input file.

Elliptical interpolation
When ODEON translates the polar input file it has to interpolate values between the four polar
planes given in the input data. This is done using elliptical interpolation independently for each
frequency band, creating the 8 x 4 plots missing between the four input plots.

An example: Polar_Omni.dat The polar input format can be found in the

DirFiles\So8_ASCII_input_file_samples\ directory, created at the installation of ODEON.

When the complete directivity characteristics are known (FULL)

The first non-comment line of the file should start with the word FULL. In the full case there are 36
lines of data for each frequency. The first 36 lines are for 63 Hz, the next 36 lines for 125 Hz, and so
on.
As a minimum there must be 1 + 36 * 8 lines in a full input file.
• 1st line is vertical upper plot 0° (12 o'clock plot, when looking from the front of the source
towards it, e.g. at a loudspeaker membrane).
• 10th line is horizontal left plot 90° (9 o'clock plot).
• 19th line is lower vertical plot 180° (6 o'clock plot).
• 28th line is right horizontal plot 270° (3 o'clock plot).

An example; Full_Omni.dat The full input format can be found in the DirFiles\So8_ASCII_input_file_samples\
directory, created at the installation of ODEON.

When the directivity pattern is rotationally symmetric (SYMMETRIC)

The first non-comment line of the file should start with the word SYMMETRIC. In the symmetric
case there is one line of data for each frequency.

As a minimum there must be 1 + 8 lines in a symmetric input file. The symmetric sources could be
a Trumpet or the Omni directional source. An example; Symmetric_Omni.dat on the symmetric input
format can be found in the DirFiles\So8_ASCII_input_file_samples\ -directory, created at the installation of
ODEON.

55
Creating a new directivity pattern using a text file as input
Once the text input file has been created in one of the formats specified above (e.g. in the ODEON
editor; ODEONEdit), it can be translated into an ODEON Directivity file, which can be applied to any
point source from within ODEON.

To translate the created text file into an ODEON directivity file:

• Select Tools|Create directivity (.So8) from ASCII file (.DAT).
• Open the input file you have created.
• Specify the name of the directivity file pattern you wish to create.
• Select whether you wish a Calibrated source or not.
• Apply calibration data as prompted for.

Applying Calibration
When creating a new directivity file you will be asked whether to create a calibrated source or not:

Calibrated source (Sound Power Level = 0 dB re 10E-12 W at 1 kHz) ? [YES / NO]

Calibrated Sources
Press [YES] if a generic source with an adjustable level is needed for ODEON-calculations.
An example on this could be the OMNI or SEMI directional directivity pattern. When selecting a
calibrated source, no data apart from the ASCII input file are required. The directivity represented
by the text-file is preserved, but the values are simply shifted by a constant amount (the same for
all bands), so the sound power level of the source is 0 dB re 10-12 Watts at 1 kHz. Please do note that
the power in the other bands may differ from 0 dB. You may still alter the overall power response
of the source by applying an EQ, however the power at 1 kHz will always end up as 0 dB, and the
other bands shifted accordingly.

NON-calibrated sources – Electro acoustical sources, machinery, natural sources etc.

Press [NO] to preserve the sensitivity of an electro acoustical source or the absolute level of natural
source, e.g. a human voice. When selecting the NON-CALIBRATED source you are allowed to enter
equalising, electric losses (zero for natural sources) and sensitivity at a selected frequency band
(zero for natural sources).

The addition of electrical sensitivity, electrical input power and electrical loss values completes the
data necessary to generate a source directivity file directly readable by ODEON 3 or newer.

Other directivity text import formats

Odeon supports the text import format used as input for generating binary CLF files. The format
is specified in the CLF Trial Authoring kit available at http://clfgroup.org/clfdocuments.htm. You
will though probably not write data in this format yourself, but it may be used for importing data
exported from other software such as EASE SpeakerLab Pro. Odeon will store the imported data
in its own native format .SoCF1Tab and .SoCF2Tab – unlike the CLF data these data are not
traceable therefore encourage that you only make use of these files for internal use and that you
do not distribute them to third party. Similar do not make use of these files from an external source.

56
 Materials
When the room geometry has been created and assigned sources and receivers, a next important
step is to assign the acoustic materials to the surfaces of the geometry.
The Material list is described in short in the Quick Start Guide (available using the Ctrl+F1 shortcut
from within the Odeon software and online at https://odeon.dk/learn/quick-start-guides/). This
manual will not repeat the content of the Quick Start Guide, so please read that first.

4.1 Overview of the Material list

The contains two lists, the Surface list on the left and the Material library on the right. The
Material list
is connected to the 3D Materials window; when selecting a surface in the Surface list, the
Material list
surface is automatically highlighted in the 3D Materials window.

The Surface list and the Material library have different functions.

• The Surface list is stored with every room, as a .mat file. It contains all the materials assigned
to the surfaces in the room, as well as any modifications made on their absorption coefficients.
These are the active absorption coefficients which will be part of the calculations. They are not
affected by any changes made to the Material library.

• The Global material library is the default library of materials available in ODEON. This
library, and any changes you make to it, will be available every time you start a project on
your computer. The Global library is stored with ODEON on your PC (by default, as
Material.Li8), so if you make changes to it, it will be different from the Global library on another
computer. This does not prevent projects from being sent back and forth between colleagues,
because the .mat file is not affected by changes to the library file.

57
• Additionally, you can make a Room material library. If you press the Toggle between global and
local room library button, you will create a material library for the room you are in. This library
file is saved as a .li8 file with the room’s name, and it will only be active for that specific room.
This is good as a sandbox, in order to avoid messing up the Global library when working.

You can change from local to global as you please, the Surface List will not be affected by any change
unless you actively change the assigned materials.

Surface list (Left side of the Material list)

In the Surface list, you will find the material specifications assigned to the surfaces. The list consists
of either all surfaces in the room, or a subset of chosen layers if using the Set Layers option (Ctrl+L
shortcut).

Surface list description

Optional description of the surface list. This is useful if you wish to study different room
configurations. You can save up to 10 versions of the Surface list, and the description can help you
to recognize them. See more in Sec.4.4.

Surface number
The Surface number is a unique number assigned to each surface in the geometry file. ODEON uses
the surface number for attaching the correct material data (and grid surfaces, reflector surfaces etc.
elsewhere in the software). When importing the geometry from SketchUp with the SU2Odeon plugin,
the numbers are generated and stored in the SketchUp model. This ensures that numbers stay
consistent even when making changes to the SketchUp model and exporting it again. When
importing a geometry e.g., in the .dwg, .dxf and .3ds file formats, the surface numbers will be
generated by ODEON. The numbering may not be the same if the geometry is modified and
reimported (e.g., deleting, adding or changing some surfaces). Therefore, when reimporting
geometry, you should ensure that data such as material properties are still correct.

Material number
The Material number corresponds to the material assigned to the surface, as listed in the Material
library. There can be differences between the material in the Surface list and the corresponding one
in the Material library, if it has been modified in either of the two lists and it has not been
reassigned.

58
Scattering coefficient
The Scattering coefficient determines the amount of scattering introduced by the surface, due to its
roughness profile. It is assigned per surface, not per material. This is because scattering depends
on the geometrical structure of the surface, and not the absorption coefficients. Two surfaces with
the same material can thus have different scattering properties. See more in Sec. 4.6. The default
value is 0.01 (1%) and it is indicated in yellow if you have not changed it.

Transparency
The Transparency coefficient can be used for surfaces with openings, which could let sound travel
through without being absorbed. Then, these openings do not have to be modelled explicitly. For
example, a layer of blinds being 30% open can be represented as one surface with 0.3 transparency.
By default, surfaces are not transparent (transparency 0%). See more about transparency in Sec. 4.7.

Remember: Transparency should not be used for transmission calculations. See more about
Transmission in Appendix B: Specify Transmission Though Walls.

Type
Odeon contains four Surface types:
• Normal: default option, to be used in most cases.
• Exterior: the surface is considered as an exterior surface.
• Fractional: the surface is a part of a larger group of surfaces.
• Transmission: to use when sound is transmitted through the surface to another room.
See more in Sec.4.8.

Surface name
The Surface name is only used in models which were built manually in a text editor (see Appendix
G: Modelling rooms in the ODEON Editor). In most cases when the model is imported, the Surface
name is the same as the Layer name.

Layer
This column shows the name of the layer which contains the surface in the SketchUp model, or in
your CAD application if the model is imported in the .dwg or .dxf file formats. For .3ds and .obj file,
the layer name corresponds to the name of the material used in the CAD software (e.g., in 3D Studio
Max).

Area
This is the calculated surface area of the surface. Occasionally, a surface can get a wrongly
calculated area which can even turn out to be 0. This will be the case if the edges of the surface are
crossing each other e.g., for a so-called butterfly surface. It is always a good idea to simplify the
model and eliminate such surfaces.

Absorption table
The table at the bottom of the Surface list shows the absorption coefficients of the surface in octave
bands. The coefficients can be edited in the table. This will update the coefficients of all surfaces
which have been assigned this material. However, changes do not affect the corresponding
material in the Material library.

Material library (Right side of the Material List)

The Material library is a list of materials that you can use for your projects. You can add new
materials, delete existing materials and change them as needed (see Sec.4.3). The materials in the

59
list are either defined by a set of absorption coefficients per octave bands or by an underlying
construction model (via the Material calculator).

Number
Unique identifier for the material. ODEON follows a numbering system which is explained further
in Sec. 4.2.

Specification
Description of the material and reference to documentation. This reference can be a link to the
manufacturer’s internet page or the measurement report written. URLs can be written as
http://odeon.dk or www.odeon.dk, and they are displayed as active links. Clicking the link will
open the specified page in the default internet browser.

Calculated materials
If the material is derived from the Material calculator, the line background is blue. Otherwise, if it
is simply specified by absorption coefficient values, the line background is white.

Absorption table
The absorption coefficients per octave band are displayed in the lower table, as well as the weighted
absorption coefficient 𝛼(𝑤) which is a broadband parameter, and the absorption class. The last
column “Model” indicates whether the material has been created with the Material calculator. As
opposed to the table in the Surface list, the table cannot be edited directly in the Material list.

4.2 Categories in the Material library

By default, the Material library follows a numbering system which you can use to keep it organised.

The default material database is divided into different categories (e.g., gypsum, wood, etc.). Most
of the categories span a range of 1,000 materials (except for the first two). In each category, the first
100 materials should be reference materials. ODEON has suggested a few to begin with. Hereafter
it will be possible to include 9 different manufacturer lists (e.g., one manufacturer with numbers
within 1100-1199 and another within 1200-1299). Following this convention, the leftmost digit
defines the type, the next one the manufacturer and the last two the specific material. Some of the
materials come from (Harris, 1991), while others have the reference and link included in the
Material description, inside ODEON.

60
The numbering of the main categories is as below:
• 0-99 Special materials.
• 100-999 Concrete.
• 1000-1999 Brick.
• 2000-2999 Ceramics.
• 3000-3999 Wood.
• 4000-4999 Gypsum.
• 5000-5999 Steel.
• 6000-6999 Vinyl/Plastic/Plexiglas.
• 7000-7999 Carpets.
• 8000-8999 Curtains and blinds.
• 9000-9999 Natural materials (e.g. sand, grass, water, etc.).
• 10000-10999 Doors/Windows/furniture/inventory (e.g. organ pipes, ventilation grills,
bookshelves).
• 11000-11999 Audience areas +/- people.
• 12000-12999 Mineral wool.
• 13000-13999 Wood wool and alternative porous absorbers.
• 14000-14999 Slit absorbers/ Micro-perforated absorbers/ miscellaneous.
• 15000-15999 Examples of materials from the Material calculator.

Special materials
There are three special materials in the library. Special materials cannot be modified.

Material 0 (transparent)
Assigning Material 0 to a surface corresponds to removing the effect of the surface completely from
all calculations. Therefore, surfaces with this material assigned:
• Offer no hindrance to rays, neither in energy or direction.
• Are excluded from the calculated active surface area of the room, and therefore do not affect
the estimate of the room's volume produced by Global Estimate or Quick Estimate
Reverberation.

This feature can be used to temporarily "remove" surfaces such as doors or reflectors from the room
- or to define a phantom surface which has no effect on the room acoustics simulation but over
which an energy map (a grid) can be plotted.

Material 1 (totally absorbing)

The totally absorbent material (Material 1) can be used for modelling outdoor situations, e.g., an
open roof. This is the only material which will stop the rays during ray tracing. No reflections will
be generated from surfaces assigned this material. Using this material when some surfaces are
known to be 100% absorbing can speed up calculations considerably.

Material 2 (totally reflecting)

Assigning Material 2 (totally reflecting material) on a surface makes it fully reflective. This is not a
realistic material, as you should expect at least some absorption from any surface, but it can be
useful to investigate a model and its limitations, as an extreme case.

4.3 Editing/extending the Material library

You can edit the Material library in multiple ways. The following buttons will open a Material
Editor with different functions to enter absorption data:
• Add material (Ctrl+M) copies the selected material in the Material library.
• Edit a Material (Ctrl+E) modifies the selected material in the Material library.

61
The material’s specifications can be edited in the Material Editor. The material number cannot be
modified, but if you create a new material, you can specify a number of your choice, provided that
it is not already used in the library.

You can also edit the library file (.li8) in a text editor, where it is easier to make multiple changes to
materials. Finally, materials can be deleted via the Delete button.

These options can be used both for the local and the global library. The changes will not be applied
to the assigned materials in the Surface list ( .mat file). Remember that material 0, 1 and 2 must
remain as originally defined.

Define a material

Enter the absorption coefficient directly per octave band in the table. The values should be between
0 and 1, or entered as percentage.

Mix materials

Create a new material by mixing several existing materials. This can be useful for surfaces which
are composed of different materials.

Max absorption of…

Create a new material by taking the largest absorption coefficient out of two materials in each
octave band. This tool can help to model surfaces that present different absorption phenomena in
different frequency ranges. For example, a carpet on a wooden floor would include high-frequency
absorption due to the carpet and low-frequency absorption where the floor behaves as a
membrane.

Paste materials from Excel (or similar software)

If you have many materials, or you have an old list of materials made in Excel or similar, Excel
sheets can be used to import it to Odeon. More information can be found on
https://odeon.dk/downloads/materials/acoustic-absorption-data/.

Material calculator
Create new materials from a series of material layers and in specified mounting conditions. The
Material calculator is explained more in detail in Sec.4.5.

62
Modifying .li8 files
You can also modify the Material library in a text editor. The Edit Material library file button opens the
.li8 file directly in the ODEON text editor. The data format in a .li8 file is very simple; each material
is described by two lines:

ID_Number Descriptive_text
α63 α125 α250 α500 α1k α2k α4k α8K

ID_Number must be a unique number between 0 and 2.147.483.647.

Descriptive_text contains the material’s specifications, as they appear in the table of the Material list.

The absorption coefficient α for each octave band should have a value between 0.00 and 1.00. If no data
is available for upper or lower bands, we recommend using the value of the neighbouring band
(e.g., use the value in the 4 kHz band for the 8 kHz band as well).

Caution! If the material is derived from the Material calculator, the message {Material Calculator. Do not
edit manually!} is displayed in the descriptive text, and we recommend not to edit it. Otherwise, you may
corrupt the file.

4.4 Editing the Surface list

The Surface list contains the material properties used in the ODEON simulation. In terms of
absorption, most of the data comes from the Material library, but the absorption coefficients can
also be edited in the Surface list.

Assign materials from the Material library

ODEON offers four functions to assign a material from the Material library to surfaces:
• Assign material: the selected material is assigned to the selected surface(s) in the Surface
list (use Shift+Left mouse button to select several surfaces).
• Global replacement: the selected material is assigned to all surfaces that have the same
material as the selected surface.
• Global assign to layer: the selected material is assigned to all surfaces belonging to the
layer of the selected surface.
• Assign material to all visible surfaces: the selected material is assigned to all visible
surfaces in the Surface list. If only some layers are displayed (as specified in the Set Layers
menu), then the material is only assigned to the surfaces of the active layers.

Note: you can also assign a material by typing its number in the Material column.

Modify materials in the Surface list

Once materials have been assigned to surfaces, it is still possible to edit the absorption coefficient
directly in the Surface list. Select a surface and edit the absorption table at the bottom of the list.
Values can either be between 0 and 1 or entered in percent.

The modifications apply to all surfaces which have the same material as the selected surface and
they are saved in the room’s .mat file. However, these changes do not affect the Material library.
This results in a mismatch between the material’s properties in the Surface list and the Material
library.

63
 Note: as a general rule, the materials indicated in the Surface list correspond to the ones in the Material
library. However, differences can occur between the two, when:
• The material has been edited in the Material library after the material was assigned to the surface.
• The absorption data of the material has been edited in the Surface List.
• The material was assigned on another computer, using another Material library with different
material definitions.

This mismatch does not affect the project, as the relevant material data is saved with the room project
(as a .mat file).

If you want to apply changes in the Material library to your project, remember to reassign the materials
afterwards (for example, using Global replacement). After reassignment, the material in the .mat file and
the .li8 file will match.

Other surface parameters

Surfaces are assigned other properties, including the Scattering coefficient, the Transparency
coefficient, and the Wall Type. These properties are all independent of the absorption coefficient,
so they are not linked to the Material library. Moreover, they are specific to a surface – not a
material, as two surfaces with the same material can have different properties.

Material list archive

Since version 16, ODEON saves up to 10 previous versions of the Material list, which you can
browse through using the Material list archive button.

The archive can be useful to switch between different material configurations of the same room
and keep track of the different versions. Typical examples are investigating different materials to
the same surface, or studying the effect of adding/removing a wall in the geometry. For each
version, you can update the description either directly in the Surface list or in the Material archive
menu.

To save a current configuration, use the Save icon. Note that until you modify it, the current version
does not appear in the Material archive.

Caution! Only one set of simulation results is saved at a time, to keep it consistent with the material list.
Therefore, changing the Material list version will delete any calculated results. If you want to keep the
simulation results of different versions, use multiple copies of the room instead.

64
4.5 Material calculator
(Combined and Auditorium editions only)

With the Material calculator, you can create and customize constructions made from different material
layers (e.g., perforated plates, porous layers or membranes). This tool can be useful to design
absorbers, for instance by varying the geometry of a perforated plate or adjusting an air cavity
behind a panel. It can also be considered a pedagogical tool to study absorption mechanisms in a
given construction.

Backing
The material construction includes a backing, which is the termination of the construction. It is
selected in the first line of the table. Two options are available:
• If the construction is placed directly on a wall, choose Rigid backing.
• If the construction is not backed by a wall, choose Air backing.

Material layers
The Material calculator is based on the transfer matrix method, in which the material is represented
by a transfer matrix, relating the pressure and the normal particle velocity on either side. The
material is built as a stack of material layers, each represented by its own transfer matrix. The
construction’s transfer matrix is obtained as the product of the material layers’ transfer matrices
(Allard, et al., 2009).

Two different models are used, according to the material layer:

• Equivalent fluid layers are layers in which the sound propagation from one side to the other
is effectively modelled. The transfer matrix of equivalent fluid layers has the following
structure,

𝑍𝐶
cosh(𝛾 cos 𝜃 ℎ) sinh(𝛾 cos 𝜃 ℎ)
cos 𝜃
𝑀𝑙 = (cos 𝜃 ),
sinh(𝛾 cos 𝜃 ℎ) cosh(𝛾 cos 𝜃 ℎ)
𝑍𝐶

65
 where ℎ is the thickness, 𝑍𝐶 is the characteristic impedance and 𝛾 is the propagation constant.
𝜃 is the angle of propagation in the layer, and it is added to account for the angle dependence
of the results.
• Lumped layers correspond to an additive impedance term. The lumped layers are
characterized by a single quantity, which we call layer impedance 𝑍𝐿 . The transfer matrix is

1 𝑍𝐿
𝑀𝑙 = ( ).
0 1

It represents the following relation between the sound field at the front and at the back of the
layer,
𝑝𝑓𝑟𝑜𝑛𝑡 = 𝑝𝑏𝑎𝑐𝑘 + 𝑍𝐿 𝑢𝑏𝑎𝑐𝑘 ,
𝑢𝑓𝑟𝑜𝑛𝑡 = 𝑢𝑏𝑎𝑐𝑘 .

In general, lumped layers cannot be placed directly in front of a rigid backing, for which 𝑍𝑏𝑎𝑐𝑘 =
∞, because in that case the calculation is invalid.

Each material layer contains a set of parameters, which can be edited below the table. The name,
the layer type and the thickness can be changed in the table directly. We will now cover the
different material layers available in the Material calculator.

Fluid layer
The fluid layer represents a lossless medium, with the following input parameters:
• Medium density (kg.m-3),
• Speed of sound in the medium (m.s-1)

Pre-sets are available for air and water. The air pre-set is particularly useful to model cavities in the
material construction.

Note: an air layer at the front or in the back of the construction does not change the absorption
properties of the material. Therefore, using an air layer at either termination is not recommended. If
you use an air backing, the first layer should not be an air layer.

Porous layer
Porous layers can be used to model porous materials such as mineral wool, foam, etc. They are
equivalent fluid layers, which follow Miki’s model (Miki, 1990). Miki’s model is based on one input,
which is the material’s flow resistivity (N.s.m-4).

Although the flow resistivity can be measured, it is not always available in a material’s
specifications. Nevertheless, you can find empirical models which relate the flow resistivity to
other material properties. The example below can be used to characterize wool-like materials, and
it relates the flow resistivity to the material’s density and the diameter of its fibers (Bies, et al., 1980).

66
 1E+06

Flow resistivity [N.s.m-4]

1E+05
1 μm
2 μm
5 μm
1E+04
10 μm
20 μm

1E+03
1 10 100 1000
Density [kg.m-3]

Fibrous materials: empirical relation between flow resistivity and density for different fibre
diameters. From (Bies, et al., 1980).
The fibre diameters depend on the material. Typical values can be found in the literature
(Christensen, et al., 1993):
• Glass wool: 1-8 μm;
• Rock wool: 2-5 μm;
• Slag wool: 2-5 μm.

Perforated plate
This category represents rigid plates with circular holes.
We assume the holes are arranged as squares, as shown
on the right figure.
The plate follows an equivalent fluid model from
Johnson-Champoux-Allard (JCA) (Jaouen, et al., 2011).
The input parameters of this model are the following:
• Hole diameter (mm).
• Perforation ratio: ratio between the hole area and the
plate area, number between 0 and 1.
As sound is only transmitted through the holes, we do
not include angle-dependence in the model. An end-
correction is also added on both sides of the plate, in
order to account for radiation effects at the outlets of the
holes.

Some aspects to be aware of:

• The plate is assumed to be made of a perfectly reflective material, so it has no effect on the
absorption properties. Similarly, bending is not included.
• The model only calculates absorption, not scattering. For small holes, we do not expect a too
large scattering effect from the perforated plate.
𝜋
• The perforation ratio is limited to (approximately 0.79), otherwise the holes overlap.
4

67
Perforated plate with rectangular holes
This material layer represents a plate with rectangular slots or infinite slits, following an equivalent
fluid model by (Vigran, 2014). Just like the perforated plate, we disregard angle-dependence and
we add an end-correction on either side of the plate.

The input parameters are the following:

• Infinite length: this makes it possible to choose between rectangle slots or infinitely long slits.
When checked, the length parameters are deactivated.
• Slot width/slot length: slot dimensions (mm). The slot length is deactivated for infinite slits.
• C-C width distance/ C-C length distance: center-to-center distances between the holes (mm),
in two dimensions. The C-C length distance is deactivated for infinite slits.
The perforation ratio is automatically calculated and it is displayed as a number between 0 and 1.

Some remarks:
• Like the perforated plate, we assume that the plate has no effect on absorption, and that it is
made of a totally reflecting material.
• The model does not include scattering, only absorption.
• The slot dimensions must remain smaller than the respective C-C distances, otherwise the holes
overlap.

Membrane
The membrane type can be used to represent thin impervious sheets, such as plastic or rubber
coverings, or plates such as a gypsum board. The model is based on a formula from (Rindel, 2018).
It includes angle-dependence and the size of the plate. It is a lumped layer, so you cannot place it
directly on top of a rigid backing.

The input parameters of the model are as follows:

• Density (kg.m-3).
• Young’s modulus (GPa).

Advanced parameters:
• Poisson’s ratio.
• Internal loss factor.

In general, you don’t need to change the advanced parameters – as they are set to their most
common values.

Typical materials can be found in the table below (Rindel, 2018).

Density Young Poisson ratio Internal loss
(kg/m3) modulus factor
(∙109 N/m2)
Gypsum board 840 3 0.30 0.015
Glass 2500 60 0.24 0.001
Steel 7800 210 0.33 0.0001
Aluminium 2700 72 0.33 0.0001
Wooden chipboard 700 4.6 0.20 0.02

Note:
• Membranes cannot be placed directly in front of a rigid backing.
• When modelling a plate, we assume it has a rectangular shape, of the same dimensions as the
entire panel (see next section). Therefore, the resonance of several smaller panels placed next to
each other is not included in the model.

68
Panel dimensions
The dimensions of the panel can either be infinite or finite – assuming a rectangular size. Finite-
sized samples include diffraction effects at the edge of the sample.

Note: for finite panels, we always assume that the material is surrounded by a rigid baffle. If Air
backing is selected, this means that the construction is built on an aperture in the baffle, see below.

Plotted results
On the right side of the calculator, the following quantities can be plotted in a graph:
• Layer properties: these depend on the model of the layer.
o for equivalent fluid layers (fluid layer, porous layer, perforated plate, plate with rectangle
slots): characteristic impedance and propagation constant.
o for lumped layers (membrane): layer impedance.
• Surface impedance of the construction. We assume extended reaction so the surface impedance
is angle-dependent.
• Radiation impedance of the panel. For an infinite panel, the radiation impedance is constant
1
and equal to , 𝜃 being the angle of incidence. Finite-sized panels are assumed to be
𝑐𝑜𝑠(𝜃)
rectangular, and the radiation impedance is calculated following (Davy, et al., 2014).
• Reflection factor of the construction. It is derived from the surface impedance and the radiation
impedance as
𝑍𝑆 − 𝑍𝑅∗
𝑅= .
𝑍𝑆 + 𝑍𝑅

• Angle-dependent absorption coefficient of the construction, derived from the reflection factor,

4𝑅𝑒(𝑍𝑆 )𝑅𝑒(𝑍𝑅 )
𝛼 = 1 − |𝑅|2 = .
|𝑍𝑆 + 𝑍𝑅 |2

The full-octave averaged values and the third-octave averaged values are also displayed.
• Random absorption coefficient of the construction. For finite-sized samples, the displayed
absorption coefficient recreates the measurement conditions in a reverberation chamber (ISO
354, 2003), with the following formula,,

𝜋
2 8𝑅𝑒(𝑍𝑆 ) 𝑅𝑒(𝑍𝑅 )2
𝛼𝑟𝑎𝑛𝑑𝑜𝑚 = ∫ sin(𝜃) 𝑑𝜃.
0 |𝑍𝑆 + 𝑍𝑅 |2 |𝑍𝑅 |2

This absorption coefficient can reach values larger than 1.1 Both the full-octave averaged values
and the third-octave averaged values are displayed. The octave-averaged random absorption
coefficient is the one used by ODEON in calculations.

The complex quantities can either be displayed as a function of frequency or as a Nyquist graph
(imaginary part as a function of real part). This is particularly helpful for the surface impedance
and the reflection factor. The plot can be displayed both in linear and in logarithmic scales. All
impedances are normalized to the impedance of air.

1
The formula is based on the statement that maximum absorption occurs for an open window of the dimensions of the
panel. It is the subject of a future publication under writing.

69
Incidence angle
The incidence angle can be varied for angle-dependent quantities (surface impedance, radiation
impedance, reflection factor, oblique absorption coefficient).

Note: changing the incidence angle does not affect the random absorption coefficient, which assumes
a diffuse field incidence.

Exporting data
The following options are available to export data from the context-sensitive menu :

• Copy absorption thumbnail: copy the exported absorption coefficient in octave bands
as an image.
• Copy selected graph to clipboard (Ctrl+C): copy the graph as displayed in the
calculator.
• Copy values to clipboard (Ctrl+E): copy the values displayed in the right-side graph as a
function of frequency.
• Copy 1/3 octave values to clipboard (Ctrl+Shift+E): copy the values in third-octaves.

Adding a material to the material list

Once the construction is ready to be used as a material in ODEON, click on Transfer absorption data to
material. This will update the bottom part of the window: the absorption coefficient is filled in from
the octave-averaged random absorption coefficient.

Although any sample size can be specified in the Material calculator, we recommend using
sufficiently large areas (above 10 m 2) or even infinite samples when transferring the absorption
data to the material. Smaller samples will be assigned higher absorption because of edge
diffraction. However, the reflection-based scattering algorithm of ODEON (see Chapter 9:
Calculation Principles) will apply extra absorption to these surfaces during a simulation. Therefore,

70
it is wise to avoid adding extra absorption due to edge diffraction twice, by using a sufficiently
large area. A warning message is displayed if the material’s area is smaller than 10 m2. If the
absorption coefficient is higher than 1, it is truncated to 1, following the methodology of (ISO 354,
2003).

A warning message is also displayed if some absorption values are extremely low (<0.01). Indeed,
highly reflective surfaces may be unrealistic. This leaves you two options:
• Overwrite the data: absorption values below 0.01 are replaced by 0.01, in order to include at
least some amount of absorption.
• Keep the data: the absorption values are kept as calculated, even the very low values.

In any case, you can freely change the values of the absorption coefficient after transferring the
material, for instance if some of the values are known or if you want to ensure another minimum
absorption coefficient.

A series of indications are displayed in the absorption table of the Material editor.

• Low absorption values: indicates absorption coefficients below 0.05. Too reflective surfaces
might be unrealistic because any surface absorb at least a fraction of acoustic energy.
• High absorption values: indicates absorption coefficients above 0.95. Too absorptive surfaces
are also rarely found in real life.
• Manually edited values: indicates the absorption coefficients that differ from the calculated
values in the Material calculator. This happens when you have modified the absorption
coefficient after transferring the material, or if values have been truncated during the transfer
(𝛼 > 1 or 𝛼 < 0.01).

You do not need to change the highlighted values; the warnings simply highlight potential issues
in the designed absorber.

The button Remove construction on the right of the material name deletes the underlying model. You
can use it if you want to keep the absorption coefficients without the calculator part.

4.6 Scattering coefficient

While the absorption coefficient controls the amount of sound energy that is absorbed from a surface,
the scattering coefficient determines the way sound energy is reflected. The basics will be explained
below. For a deeper understanding of scattering coefficients, please refer to Chapter 9(Calculation
Principles).

We can group reflections in rooms in three categories:

• Specular reflections: where the angle of reflection is equal to the angle of incidence.
• Diffuse reflections: where the angle of reflection is independent of the angle of incidence
(memoryless reflection).
• A mixture of the cases above.

71
By default, ODEON uses a mixture, including a specular part and a diffuse part to sound
reflections. Scattering is modelled in two ways in ODEON:
• Surface scattering, which is quantified by a scattering coefficient, assigned to the surfaces in the
Surface list.
• Scattering due to diffraction, which is included if Reflection Based Scattering is enabled in the Room
Setup.
These two scattering processes are described more in detail in Chapter 9 (Calculation principles).

Note: You can deactivate Surface Scattering and Reflection Based Scattering in the Room setup, but this is only
recommended for educational purposes and research.

As explained in Chapter 2, the room’s geometrical model may need to be simplified for room
acoustic simulations. In that sense, many fine details such as rough surfaces, objects placed on the
walls, windows and crowds will be represented by flat surfaces in the ODEON model. A model of
surface scattering is thus added to ODEON to account for scattering by these elements.

The scattering coefficient is a number between 0 and 1 which represents the ratio of non-specular
reflected energy to the total reflected energy. It is a frequency dependent quantity but ODEON
uses a single number, which corresponds to the scattering coefficient at mid-frequencies (707 Hz).
By default, ODEON will expand the coefficient to appropriate values for the other frequency bands,
using a typical behaviour of scattering in relation to frequency. This single value scattering
coefficient is assigned to each surface in the Surface List.

Note: Some acousticians call the scattering coefficient diffusion coefficient, but this is wrong. The
diffusion coefficient of a surface describes the spatial uniformity of reflections from it, while the
scattering coefficient determines the fraction of sound which not reflected in the specular way. A high
scattering coefficient can lead to a high diffusion coefficient, but the two values are not directly linked.
In addition, they are measured in two different ways (ISO 17497-1, 2004) and (ISO 17497-2, 2012).

Suggested values for the scattering coefficient

Remember: As a user you can only assign one value of scattering coefficient per surface. This value
corresponds to 707 Hz (the centre frequency of the two mid-frequencies 500 and 1000 Hz, in a
logarithmic scale). However, ODEON gives different weighting for the rest frequencies, according to the
curves in Sec. 9.5: lower frequencies have less scattering, while higher frequencies have higher
scattering.

In ODEON, the default scattering coefficient is 0.01 (1 %), because real surfaces are never entirely
smooth. A scattering coefficient of 1% is still on the low side, and it will still highlight potential
acoustic problems such as flutter echoes. The reason to start with a value of just 1% is that we
should first highlight these acoustic problems, and then actively choose to increase the scattering
coefficient. When the scattering coefficient is left at its original value of 1%, it is highlighted in light
yellow, indicating that you may consider putting your own value. A scattering coefficient of 2 - 5%
for a smooth surface is generally a realistic value. The scattering coefficient should be higher for
surfaces which have lost some level of detail in the geometrical model. Even a few rough objects
sparsely placed on a surface greatly increase the scattering properties of the surface.

72
The scattering coefficient (at the mid-frequency 707 Hz) that should be used for various materials
is an ongoing research area. Our current recommendations for real structures are the following:

Material Scattering coefficient at mid-frequency (707 Hz)

Audience area 0.6–0.7
Rough building structures, 0.3–0.5 m deep 0.4–0.5
Coffered ceiling 0.3-0.4
Bookshelf, with some books 0.3
Brickwork with open joints 0.1–0.2
Brickwork, filled joints but not plastered 0.05–0.1
Smooth surfaces, general 0.02–0.05
Smooth painted concrete 0.005–0.02
Suggested scattering coefficients to use for various materials. The given values are to be assigned to surfaces in
ODEON and correspond to the middle frequencies at 707 Hz. Suggestions may be subject to changes as more
knowledge on the subject is achieved.

With the suggested scattering coefficients, it is assumed that Reflection based scatter has been enabled
in the Room setup, in which case ODEON will account for effects such as distance, angle of incidence
and edge diffraction (read Chapter 9 about Calculation Principles for more information). If Reflection
based scatter is disabled, then it is recommended to increase the scattering coefficient further (default
scattering coefficient to 0.1; 0.3 may be more appropriate for disproportionate rooms such as class
rooms).

Example: diffuser with pyramidal wedges

The example below can serve as a rule of thumb to estimate the scattering coefficient of a rough
surface, following a theoretical model based on (Rindel, 2015). We consider a surface with a cross-
section as shown in the figure below. The surface has a pyramidal structure which should be
represented as a flat surface in the simulation model. Scattering will then be accounted for by
assigning a scattering coefficient to the surface.

2d

A diffuser with a depth of structure equal to 2d . This type of structure is simplified in ODEON by modelling it as
a plane surface and assigning scattering coefficient according to the frequency functions described below.

The scattering coefficient of the surface depends on the depth of structure 2d, the graph below
illustrates this dependence according to a theoretical model (Rindel, 2015). For example, for a 1D
diffuser of depth 2d = 15 mm, the scattering coefficient is approximately 0.01 (1 %) – which
corresponds to the default value in ODEON.

73
 1

Scattering coefficient at 707 Hz

0.1 1D Diffuser
2D Diffuser

0.01
10 100
Depth of structure 2d (mm)

Scattering coefficient of a surface with a pyramidal diffuser according to the depth of structure 2d (Rindel, 2015).

As a first approximation, the scattering coefficient of an arbitrary rough surface can be read from
this graph by using an average depth of structure. However, you should remember that this is
based on a theoretical model and a specific roughness profile, so the actual scattering coefficient
might differ.

4.7 Transparency coefficient – semi-transparent

surfaces
A transparency coefficient can be assigned to each surface; this is a way to make the surface semi-
transparent. This feature may be used when modelling many small surfaces grouped together, in
a simplified manner. For instance, a reflector panel built from many small surfaces with space in-
between or an installation ceiling can be modelled as one large surface having a transparency
coefficient of, e.g. 0.5. The transparency coefficient can be assigned values between 0 and 1:

• 0.0 is the default value assigned to all solid walls. This value should always be assigned to the
boundary walls of the room; otherwise, rays will escape from the model.

• Very small transparency coefficients should be avoided unless the number of rays is increased
substantially. Instead, consider modelling the surface as solid. Using a transparency coefficient
larger than zero will cause the Image source method to be discarded for rays hitting such
surfaces (only relevant for point sources). Another problem is that only very few rays will be
transmitted, making the results on the other side of the surface statistically unreliable.

• Very large transparency coefficients, e.g., 0.95 should also be avoided. Instead consider
removing the surface from the model, or more easily, assign Material 0 (transparent) to the surface.

Remember: If entering scattering and transparency coefficients between 1 and 100, ODEON will assume
that the values are in percent and divide values by 100.

The Transparency value should not be used for modelling sound transmission through walls -
instead you should use the Transmission wall type for this purpose as explained below.

74
4.8 Different Wall Types
The wall type controls what happens when a sound ray hits a surface in ODEON.

There are four different wall types in ODEON. The Normal, Exterior and Fractional types relate to the
way Reflection Based Scattering is calculated for sound reflecting from the surfaces.

• Normal is the default value. It results in default handling of scattering and diffraction taking the
Reflection Based Scattering method into account if it is enabled in the Room setup.

• forces a surface to be handled as an exterior surface even if it was not detected as such
Exterior
by ODEON. The result is that less diffraction is applied at the lowest frequencies where offset
in the wall is not sufficient to result in low frequency diffraction.

• should be used for surfaces which are fractions of a bigger whole. This could be surfaces
Fractional
being part of a curved wall or a dome. These surfaces should not cause diffraction due to their
individual area, and therefore the individual surfaces do not provide any significant edge
diffraction.
When setting the type to Fractional, the surface area used for calculating the Reflection Based
Scattering Coefficient is determined from the box subscribing the room. If the construction part
which the fractional surface is part of is considerably smaller that the 'room box', the scattering
might be underestimated and a higher scattering coefficient should be assigned to the surface.
In ODEON 17, you do not need to use the Fractional type for surfaces belonging to the same plane,
because ODEON will cluster them automatically.

• Transmission is for walls which transmit sound to another room. When this wall type has been
selected, it is possible to edit the transmission data to specify the reduction index (transmission
loss). It is also possible to link the wall with another surface in case the wall is composed of two
distinct parallel surfaces, and possibly with different surface materials on either side.
When the Transmission type is assigned to a surface, 90% of the rays hitting the surface will be
reflected and 10% will be transmitted. The energy calculations are accomplished by multiplying
each ray (or particle) with the appropriate frequency dependent energy parameters.
Transmission is covered in Appendix B: Specify Transmission Though Walls.

4.9 Material Toolbar

Below is a guide to the Material List Toolbar. For the “Repeat…” buttons described below, remember
that the scattering coefficient, transparency or wall type of a surface needs first to be entered in the
Surface list in the appropriate cell using the keyboard or mouse.

Assign Material Ins Assigns the material selected in the Material list to the
surface(s) selected in the Surface list.
Global Replacement Ctrl+G Replaces all instances of a material in the Surface list.
Global Assign to Layer Assigns the selected material to all surfaces on the
selected layer.
Assign Material for all Surfaces Ctrl+Ins Assigns the material selected in the Material list to all the
visible surfaces in the Surface list.
Repeat Scattering Coefficient for Alt+S Assigns the last scattering coefficient entered in the
Selected Surface(s)
Surface list to the current selected surface(s).
Assign Scattering Coefficient to Assigns the scattering coefficient of the currently
All Surfaces
selected surface to all surfaces.

75
Repeat Transparency Coefficient Alt+T
Assigns the last transparency coefficient entered in the
for Selected Surface(s)
Surface list to the current selected surface(s).
Edit Transmission data Alt+Y
Opens the Transmission menu. The button is only active
if the surface type is Transmission.
Repeat Wall Type Ctrl+Y
Assigns the last chosen wall type entered in the Surface
list to the current selected surface(s).
Quick Estimate Ctrl+Q
For a fast evaluation of reverberation times and a
summary of absorption areas. Not as accurate as a full
simulation, but useful for quickly experimenting with
different configurations before running a lengthier
simulation.
Add material(s) Ctrl+M Opens the material editor to insert a new material. The
new material is the copy of the currently selected one in
the Material library.
Edit a material Ctrl+E Opens the material editor to edit the currently selected
material.
Delete Material Ctrl+Del Deletes the selected material from the Material library.
Edit Material Library File Opens the Material library file in the ODEON text editor.
Reload Material Library Ctrl+R
Loads changes to the Material library file that were made
in the text editor.
Search for Surface or Material F3 Enables a search field in the Surface list or the Material
with name…
library.
Toggle interior/exterior Ctrl+T
Cycles between showing interior surfaces, exterior
surfaces, and both.
Set layers Ctrl+L
Enables or disables layers from showing up in the
Surface list and in the model.
Toggle between Global and Local Switches between the Global material library and the
Material Library
Local material library.
Limit absorption coefficient Specify a minimum and a maximum absorption
coefficient for the materials in the surface list. This is
useful because extreme values in the absorption
coefficients can have an extreme effect on results
(Oppenheim, et al., 1989).
Material list archive Alt+R Opens the Material list archive and choose from up to 10
last versions of the Surface list.

76
 Calculation results
This chapter covers the different calculation types available and the results they provide. To learn
more in detail about acoustic parameters and decay curves, please read Chapter 8 and for the
theory behind Odeon calculations read Chapter 9.
The software is made in such a way that you can easily calculate the most common results, making
it easy to get started as a new user. Using the available options from the context-sensitive menus and
keyboard shortcuts, you have access to more advanced displays covering a broad range of results.

Remember: As Described in the Quick Start Guide, the menu between Toolbar and Options always gets
updated with the active Window inside ODEON. Click that context-sensitive menu to retrieve all possible
functions and Shortcuts for the corresponding Window.
Always press F1 when you need to learn more about specific functions in ODEON. The smart help will
open to a page related to the active window. The ODEON help file complements the manual and offers
extra information.

5.1 Global Calculations

Global calculations in ODEON are the ones that do not depend on receiver positions. They still
use a source, but results refer to the room as a whole.

Quick Estimate
is a fast calculation of reverberation time which is found in the Material List, t. It is based
Quick Estimate
on analytical formulas like Sabine’s and Eyring’s, which do not take into account the actual
geometry and position of materials in the room, but only the amount of absorption and volume.

77
The tool can be used for a fast overview of acoustics in a room, while assigning materials. Several
informative tabsheets are available:
• Estimations provides fast estimation of RT based on analytical formulas, not the geometry. Only
for a first glance while working with materials. The estimated reverberation time is a good
starting point to define the impulse response length in the Room Setup . A good rule of thumb
is to set the impulse response length 2/3 of the maximum reverberation time in the Quick Estimate.
• Material Overview offers an overview of used materials, with the corresponding absorption areas
they occupy.
• Unused absorption shows which surfaces/materials have a lot of unused absorption to offer.
• Estimate area can calculate absorption area which needs to be added in order to meet a specific
target for RT.
• Alpha shows all the absorption coefficient curves for the different materials used in the room in
one graph which can be used for a quick comparison or for inclusion in a report.

Global Estimate
The Global Estimate is found in the top toolbar. It makes use of ray tracing in order to calculate the
overall/global reverberation time in the room. The tool is much more accurate than the Quick Estimate
and it depends on the selected source at the bottom drop down menu. The calculation is receiver-
independent, giving an average result for the whole room volume, while in reality there will be
some local variation though normally the reverberation time is fairly constant for a room.

5.2 Point response calculations

A point response calculation gives the response of the room for a particular receiver position (with
a given orientation). The point response is generated as a response to one or multiple sources
emitting a Dirac delta function (see vocabulary) at the same time (each source can be assigned an
individual delay). ODEON simulates the impulse response between the source(s) selected and each
receiver, out of which all Room Acoustic Parameters are calculated (see Chapter 8). Point responses
are managed from the Job List window which can be populated by up to 1000 jobs.
For each Job, the desired active sources can be selected from the available sources in the pane on
the left side of the Job List (or upper side if the Job List window size is small). Please check the Quick
Start Guide for an overview of the Job List, with examples of use.

78
Three types of calculations can be carried out with the source(s) selected for each job:

Grid Responses (I)(A)(C)

Room Acoustic Parameters (see Chapter 8) will be calculated for the surfaces selected in the Define
Grid in form of colour maps. For an overview on defining grids see Chapter 6. To calculate a defined
grid, make sure to tick both the Grid checkmark on top of the column, as well as the checkmark for
a particular Job. Click the Run Job button to run the job. Once the calculation is finished, the job
turns green.
To see a calculated Grid Response, click the calculated desired Job then click the View Grid Response button
(shortcut Alt+G).

Multi Point Reponses (B)(I)(A)(C)

Room Acoustic Parameters (see Chapter 8) will be calculated for ALL receivers defined in the Source
Receiver list. This allows to see parameters for each receiver alone or statistics for them. After point
responses have been calculated for all receivers it is possible to create groups of receivers allowing
results, statistics etc. to be displayed for a subset of the calculated receivers. To calculate a Multi-
point response, make sure to tick both the Multi checkmark on top of the column, as well as the
checkmark for a particular Job. Click the Run Job button to run the job. Once calculation is
finished, the job turns green.
To see a calculated Multi-point Response, click the calculated desired Job and then click the View
Multi point response button (shortcut Alt+M).

Available Results in the Multi Point Response

Here is a short overview of the available tabsheets in the Multi Point Response. To navigate the results,
check the functions in the context-sensitive menu bar. If needed, use the help file for additional info.
Find the help file by pressing F1 when the display of interest is selected (context sensitive).

displays the sound sources and the receiver currently active. On the right side
3D Sources and Receivers
of the tabsheet you can select groups of receivers, specify which receivers should be active in the
selected group and save the selected group for later use (remember to do so). The result displayed
in the multi point response display will be for the receivers currently active. If only one receiver is
active then some of the tabsheets listed below will not be displayed.

Energy parametersdisplays the calculated acoustic parameters in table format. The parameters could
be EDT, T20, C80 and etc. The parameters can easily be selected and copy-pasted into e.g. EXCEL if
you are interested in making graphs or statistics outside ODEON.

displays the calculated acoustic parameters as bars graphs as a function of

Energy Parameter bars (1)
frequency (octave bands). The parameters could be EDT, SPL, C80 and etc. Use the arrow keys to
navigate between parameters and the R shortcut to change Receiver.

79
Energy Parameter bars (2) displays calculated room acoustic broadband parameters. The horizontal axis
displays the receivers.

displays parameters as a function of distance, by having the receivers lined

Parameters versus distance
up from the nearest to the one furthest away (from the source).

displays statistical calculations of parameters for the selected receivers, e.g. Average,
Statistics
Standard deviation, Min/Max values.

displays parameters based on ISO 3382-3, for open plan offices, but only if the
Spatial decay curve
simulation setup is according to the standard. See application notes ISO 3382-3 Open Plan offices -
part 1 and 2 about using the standard in ODEON: https://odeon.dk/learn/application-notes/.

also displays parameters based on ISO 3382-3, parameters for open plan offices,
STI versus distance
but only if the simulation setup is according to the standard.
See application notes as above.

display measurements and simulation results for the same room you have
Measured versus Simulated
imported measured results. This is a very useful tabsheet that allow making sure that a model
matches well to reality in an acoustical renovation projects or to verify your work after a simulated
room has been built.

Single Point Responses (A)(C)

In contrast to the Multi Point response, the Single Point response derives the impulse response and Room
Acoustic Parameters for one receiver only. The Single Point response allows for more detailed results,
graphs and curves, as well as for auralisation at the receiver position (see Chapter 7). To calculate
a Single Point response, make sure to tick the Single Point Response Receiver checkmark on top of the column,
as well as to select a Single Point response receiver from the drop-down list in the Job List. Click the Run
Job button to run the job. Once calculation has finished, the job cell turns green.
To see a calculated Single Point Response, click the calculated desired Job and then click the View Single
Point Response button (shortcut Alt+P).

Available Results in the Single Point Response

Here is a short overview of the available tabsheets in the Single Point Response. To navigate the results,
check the functions in the context-sensitive menu bar. If needed, use the help file for additional info.
Find the help file by pressing F1 when the Single Point Response is open (context sensitive).

display the calculated acoustic parameters in bars graphs as a function of frequency

Parameter bars
(octave bands). Parameters could be EDT, SPL, C80 etc. Use the arrow keys to navigate within
different parameters.

Energy parametersdisplay the calculated acoustic parameters in table format. Parameters could be
EDT, T20, C80 and many more. It is easy to select, copy and pasted the parameters into e.g. EXCEL
if you are interested in making graphs or statistics outside ODEON.

Decay curves display the decay curves for each frequency band. Read about decay curves in Chapter
8.

Decay Roses display how much sound energy is received at cumulative time intervals in the
horizontal and vertical planes. This can give an idea of how different parts in the decay are formed
throughout time.

80
 display how many reflections arrive at the receiver position per millisecond as a
Reflection density
function of time. The graph can be used for judging if enough rays have been used in the
simulation. A rule of thumb is to have at least 50 reflections per millisecond.

Reflectogram displays the arrival times and

strength of Early Reflections arriving at the receiver. Time
starts counting (0 sec) from the moment the direct sound arrives at the receiver.

visualises the reflection paths corresponding to the reflections in the

3D Reflection paths/Active sources
reflectogram. Use the options from the context-sensitive 3D Reflection Paths menu to exhaust all the
possibilities of the tabsheet.

BRIRshows the Binaural Room Impulse Response. Only available if activated in the Auralisation
setup (Chapter 7). The impulse response is filtered for left and right ear, according to the HRTF
(head-related transfer function) used.

Dynamic diffusivity curves is a specialized research tool for diffusivity. Use the help file (F1) for a
thorough explanation.

is a tool which helps investigating noticeable echoes. Use the help file (F1) for a
Dietsch echo curves
thorough explanation.

Auralisation (A)(C)
Closely related to the Single Point response, the auralisation features in Odeon which makes use of the
simulated impulse response and combines it with an anechoic (dry) wave file, providing a realistic
listening experience at the receiver position. In simple words, auralisation gives you the ability to
virtually place yourself in the room and listen to what any signal sounds like at that position. Please
read Chapter 7 for a detailed tour.

5.3 3D-matrix Calculation

The 3DMatrix calculation is a special type of point response calculation which allows investigating
the response from one source position at other source positions without the need to define any
receiver positions. This makes it easy to investigate scenarios where sources by nature acts source
as well receiver. Examples on this may be the occupants in an open plan office or the musicians in
an orchestra.
The 3DMatrix utility calculates a matrix of all the point responses from each point source to all the
other point sources. So, if three-point sources are defined then nine combinations are calculated;
S1S1, S1S2, S1S3, S2S1, S2S2, S2S3, S3S1, S3 S1 and S3S3.
The responses S1S1, S2S2 and S3S3 are special in that the source and the receiver coincide.
Obviously, the direct sound in these combinations would go to infinity due to the source-receiver
distance of zero. So, in these cases ODEON excludes direct sound. These values occupy the
diagonal in the source-to-source matrix and are referred to as diagonal values. Although the
diagonal values are not values that can be physically measured, we do believe that the values can
give valuable information e.g. as to how loud the room sounds for the source’s point of view.

3DMatrix Matrix 3DDiagonal

The 3DMatrix window has 3 tabsheets:

81
3DMatrix, Matrix
and 3DDiagonal (self-response). For all the displays the selected parameter is displayed
at the selected frequency (for parameters that are frequency dependent).
In the 3DMatrix display, double click on a source to make it the active one. Its response is then
visualized at the other source positions. The Matrix display is difficult to read as there is no obvious
way to order the sources at the two axes (source/receiver) e. g. distance. However, as all values are
displayed at one graph, it is possible to spot extreme values. Double clicking a value in the graph
will be displayed in the appropriate 3D display – the 3DMatrix or the 3DDiagonal. The 3DDiagonal shows
all the diagonal values for all sources in one 3D graph.

5.4 Results from measurements

All ODEON editions include a measuring system that makes it possible to capture impulse
responses in real rooms. ODEON calculates all parameters from measured impulse responses and
allows comparison with simulations within the Multi-point response interface. For a full overview of
results from measurements see in Chapter 12.

5.5 Visualisation Tools

Visualisations in ODEON help to get the overview of sound propagation in the room model and
they form a great tool for demonstrations to clients that are not familiar with Room Acoustic
Parameters and scientific terms. Visualisations can provide great screenshots and animations for
presentations, but should not be used for accurate predictions.

3D Billiard
A number of billiard balls are emitted from a source position and gets reflected by the surfaces of
the room. This is a useful animation tool which visualizes how a burst of sound will propagate in
the model and to track specific phenomena like flutter echoes or focusing from curved geometries.
Scattering is included in the animation and to some degree it is also possible to enable visualization
on how sound gets absorbed. By default, balls change colour every time they are reflected from a
surface, as an alternative colour can be used for visualization of strength. Read more about the
different options by pressing F1 in the in this window or by seeing our tutorial videos on our
website.

3D Reflector Coverage
In the 3D Reflector coverage, you can visualize which areas a given reflector will cover, up to the
specified order of reflection. Use the Define reflector surfaces window first in order to define which
surfaces in the model will act as reflectors.

3D Investigate Rays
Rays are emitted from a source, and you can view them as the reflect from the surfaces in the room.
This is mainly a tool that can be used for checking the geometry for holes. Enable the Display sticky
lost rays only option to be able to see which rays escape from the geometry and to manage locating
leaks in your model. It can also be used for purposes like investigating how the scattering
coefficients affect backscattering in tunnels, or in what way sound is distributed in coupled spaces.

82
 Colour Grids – colour maps
6.1 Overview
The 3DGrid display displays maps room acoustics parameters in the 3d model. It is a good way to
present an acoustic parameter in the complete room or for a selected part of it using grid groups.
It can be used for displaying the variation throughout the room, to provide statistics for the defined
grid receivers and it is a useful for spotting locations of acoustical problems as well.
The colour grids option is first presented in the Quick Start Guide and the help file contains a lot of
information (press F1 from within the 3DGrid display). This chapter will give an overview and
some extra details. Grids can be defined, calculated and manipulated in four places inside ODEON:
In the Define Grid, in the Job list, in the View grid response and in the Room Acoustic Parameter list.

Define grid
These windows are where you define which surfaces you want to make grids on. The left side
contains a list of the Room surfaces. Choose to show horizontal and/or vertical surfaces from the
options at the bottom. Use the Depth column to help you locate the surfaces you are interested in.
Sort surfaces by clicking on the arrow just next to the column title. Use the hand to move
surfaces to the right side that contains the list of Selected surfaces for the grids.

You have multiple setup possibilities. Some of them are:

• Choosing the distance from the surface to the hovering grid.
• Choosing the same surface multiple times with different distance to make grids in layers.
• Choosing the distance between receivers (squares) in the grid, thus changing the resolution.
Be aware that higher resolution increases calculation time exponentially.
• Grouping the grid surfaces – which will allow display of selected parts of the grid (and
calculation of statistics of that subset of the grid receivers).
Use the context sensitive help (F1 shortcut) to learn more about the individual options.

83
 Remember: Before any calculations have actually been performed, the colours displayed in the Define
grid are random.

Joblist
The Joblist is the window where you choose for which jobs to make a grid calculation and thereby
which sources should be active. Enable the grid calculation for a Job by checking the grid checkbox
Grid for the job(s) you are interested in. To calculate jobs, click the Run Single Job or Run All Jobs button
in the right-side toolbar of the Job List. Once a grid job has been calculated its cell in the Job list will
be green, to view results click the View grid response in the right-side toolbar of the Job List.

The sound sources for the grid calculations will be those activated for the job. The receivers
however, does not relate to the job instead each grid receiver is the centre of each of the squares as
defined in the Define Grid windows.

View grid response

Once a job containing a grid has been calculated (turns green), it is ready to display grid results.
Click View grid response to see the calculated grids. Use the arrow-keys to change parameters and
frequency band.
By default, an interpolated grid (a feature introduced with ODEON 15) is displayed. Click
3D_grid>Interpolate Grid (shortcut I) to disable interpolation and see the individual calculated grid
receivers (squares).
If grid groups has been defined in the Define Grid window, then it is possible to display subsets of
the grid surfaces using the G and Shift+G shortcuts.

Remember: Disabling the Interpolate Grid option (shortcut I) will allow you to navigate between grid
results faster.

Room acoustic parameter list

Here you can change the colour, range and labels for your grids.

84
6.2 Accuracy in the Interpolated Grids
Before version 15, a grid in ODEON was just a grid of squares. Interpolation has made it possible
to obtain visually appealing maps for reports much faster, since one can achieve a very smooth
image with few grid receivers (squares). You can toggle between the normal and the interpolated
grids by choosing 3D_grid>Interpolate Grid (I shortcut) while working in the View grid response.
Even though grid interpolation looks attractive even when only a small number of grid receivers
were calculated, be aware that the actual calculation is still done for limited number of grid
receivers defined, keep in mind that the grid resolution should be high enough (and the receiver
distance low enough) to display interesting trends in the room. On the other hand, a coarser grid
will speed up the calculations and might look smoother whereas a too fine grid might give a false
idea of how accurate the model actually is. Find your own compromise between the two,
considering the room size, complexity of the geometry and calculation time.

Quite fine grid with 0.75m between receivers:

EDT:

Quite rough grid with 3m between receiver:

EDT:

85
6.3 Range, Colours and Labels for Grid Display
There are three display parameters for the grids of each acoustic parameter: range, colours and
labels.

Range
The range of the grid (min to max values) is set automatically by default for most acoustic
parameters. For STI and similar parameters the limits are by default set to a manual scale ranging
from 0 to 1. For the other parameters, the range will start around the lowest calculated value
(extreme values are excluded) and go to around the highest calculated value (extreme are
excluded). The min and max values for each parameter are the same across the octave bands.
To use a manual grid range, go to the Room acoustic parameter list and press the Expand tables button
in the upper-right corner. Check the Manual grid box for the desired parameter and set the new min
and max values. The manual ranges remain fixed between calculations, making it easier to compare
results before and after a change in the model.

Colour and Labels

By default, only a few colour scales are used: the Geography-inspired colour scale and the STI-color
scale. Particularly for the STI parameters, the colour scale is divided into different categories from
“Bad” to “Excellent”.
Colour scales can be defined for all parameters in the Room acoustic parameter list. A few of these
include labels. You can also make you own colour scale with labels from scratch, but this is a bit
more complicated. It requires to copy and edit one of the .clr files found in the C:\ProgramData\Odeon
folder. The ProgramData folder is hidden in Windows by default but you can access this ODEON
folder form the menu Options|Explore Odeon Application data folder from within ODEON.

86
 Auralisation
(Combined and Auditorium editions only)

Auralisation in ODEON makes it possible to listen to the acoustics of simulations and

measurements. It provides an experience of room acoustics through headphones or through
speakers, instead of just reading numbers. It can also be a good way to present acoustical results
for a client who is not familiar with acoustic terms and parameters.

An auralisation is a pairing of the acoustics of a room with an audio file. This way you can listen
to auralisations of speech in a canteen, an orchestra in a concert hall, disturbance from co-workers
in an open plan office etc.

Although ODEON tries to make the available auralisation features as easy as possible to use, a
chapter is needed, as this is where the threads from; room acoustics modelling, signal processing,
wave signal files, transducers, psychoacoustics and recording techniques, all meet.

In the description of auralisation, special words are frequently used. Please refer to Appendix A:
Vocabulary for a short description of frequently used words. In this chapter it is assumed that you
have tried the basic auralisation functions in the Quick start guide for ODEON Auditorium or
ODEON Combined, so it is recommended to first go through the corresponding Quick start guide.

7.1 What is Auralisation

When you make an auralisation you normally pair (convolve) an audio file without any room
acoustics (an anechoic recording) with the room acoustics of a simulation (impulse response).

• This ‘pairing’ is eventually a complicated digital signal process, and in scientific terms is called
convolution.
• Auralisations are often based on anechoic recordings, i.e. an audio recording that does not have
any reflections from a room and therefore does not include any room acoustics. The audio file
can be anything like speech, music, common sounds or noise. Situations where you would not
use an anechoic signal are if you are simulating a cinema or concert venue where the speakers
play recorded music. Then you may use normal music recordings instead.
• Auralisations made in ODEON use the impulse responses for a specific receiver position which
describes the acoustics of the room at this specific location with given source positions(s). The
direction of the receiver is also important as it represents the orientation of the listener’s ears
relative to the source.
• Auralisation in ODEON produces audio signals for playback either through headphones or a
custom surround sound system.

The Anechoic Recordings

The way you record an audio file without any room acoustics involved is perform the recording in
an anechoic environment (without any reflections from a room). This is often done in an anechoic
chamber with extremely absorptive walls and floor, where practically only the direct sound to the
microphone is present. Occasionally it can be acceptable to use a file (e.g. your own voice) recorded
in a fairly dry room or office, with a directional microphone close to the source.
Included with ODEON is a library of sound files including speech, classical instruments and more.
There is even an entire recording of an orchestra, with the instruments recorded one by one, in an
anechoic room playing various classical pieces.

87
Recorded Music Instead of Anechoic Recordings.
In the cases of auralisations/simulations of a bar, music venue, cinema or similar, where recorded
music is to be played from speakers, you do not use anechoic recordings but just a music recording
instead. In the end of the chapter there is a small guide.

The Impulse Response

An auralisation is the result of a convolution of a sound signal and an impulse response. The impulse
response is a short signal that describes the room acoustics for a specific place and direction of a
receiver in the room. There can be one or multiple sound sources generating the impulse response,
e.g. a loudspeaker system at a train platform can consist of many loudspeakers which emits the
same signal.
The acoustical calculations in the Job List are based on geometry, sound absorption, source positions,
receiver position, sound levels and more. When paired with the anechoic sound files, it will result
in a new audio file that sounds just as the original signal would sound like in the actual room (that
is, affected by the room’s acoustics), at the specific position of the receiver.
For some types of auralisations, the impulse responses are chosen from the Job List in ODEON. For
others, new jobs/impulse response are generated automatically. No matter what, there can be only
one receiver (position and orientation) associated to an auralisation.

Binaural Auralisation
For binaural auralisation (see Appendix A for vocabulary), one signal is created for each ear (bin
meaning two) – when this binaural signal is reproduced in stereo using headphones, a true 3D
experience coming close to how we experience sound in real rooms can be achieved. Only binaural
auralisation is covered in this chapter. Odeon does though also cover surround auralisation for
reproduction over loudspeakers and most of the details and principles covered here are also true
for surround auralisation – indeed for surround auralisation the aim is also to provide a binaural
experience.

Depending on whether you are interested in using headphones or a surround sound system, the
way the simulated acoustics of the room is applied to the audio file will be different.
For binaural simulations you need to make a calculation of the room acoustics received at the two
ears of a virtual dummy head (a two-channel impulse response; a channel for the left and a channel
for the right ear). This is called the BRIR (Binaural Room Impulse Response). The BRIR can be
calculated as part of the Single Point response in the Job List, using the reflection data calculated from
one or more sound sources to one single receiver. When the BRIR is convolved with an anechoic
signal and played back over headphones, it will give the same listening experience as would be
obtained in the real room.

For many Single Point response results, the direction of the receiver does not matter, but for
auralisation the direction is important. The direction defines the orientation of the simulated
dummy head that “records” the sound file when it is played in the simulated room.

Tip: If having multiple jobs in the Job List with the same single point receiver and receiver orientation
then multi-channel simulations are possible (e.g. simulating a stereo setup, background noise versus
loudspeaker announcements or singer versus orchestra).

88
7.2 Auralisation Options
When it comes to auralisations, there are different options available in ODEON. They are all
explained in more detail later in this chapter.

1. Quick Auralisations / Streaming Convolution

This is the fastest way of making an auralisation. For every job where a single point
response is calculated, you can open the Streaming Convolution to listen to different audio
files paired with the acoustics of that job. However, this type of auralisation cannot be
exported.

2. Large Auralisations / Offline Convolution

If you want to make large auralisations, with different audio files played from different
source positions you can make an offline auralisations, and/or if you want to export it as a
.wav file to show/send to a client or partner. From version 15, there is a new tool that makes
it much faster and easier to make large auralisations with many different audio files and
source positions. This is called the “Multi Source/Signal Auralisation Expert”.

3. Audio Effects
Because it can be difficult to find enough anechoic recordings to make a full crowd in a
canteen, or to add small differences on the different instruments in a classical piece, Odeon
implements an Audio Effects tool. It allows to make a single anechoic audio signal into
many slightly different audio files so you can use them together. Read more about this later
in this chapter.

Before using any of these tools, the type of auralisation and other options need to be setup.

7.3 Setup for Auralisations

The first step to make an auralisation is to define your settings in the Auralisation setup. Here you can
enable the creation of Binarual and/or Ambisonic impulse responses, which can be used in Streaming
or Offline convolution, within from the Job List.

Binaural settings
These settings concern reproduction strictly through headphones. Enable the Create binaural impulse
response file option. This will create a binaural impulse response with the calculation of any single-
point response.

89
In the Auralisation setup, the upper-half window refers to binaural reproduction, while the lower-half refers to
ambisonic reproduction (see next section).

Remember: Make sure the Create binaural impulse response file (.Jnn) option is activated when you want
ODEON to derive BRIR’s for auralisation.

Head Related Transfer functions -HRTF

For most users, the HRTF filters are not very important to learn about in detail. The HRTF file
contains the filters that mimics the effect of the head of the listener on the perceived sound. When
sound comes from left to right, the frequency response will be different on either ear and there will
be a time delay. ODEON includes this in the auralisation to make sound directions perceivable. By
default, there are two HRTF filters available in ODEON, just with the resolution being different. If
you are interested in learning more about the HRTFs or how to add new HRTF files, read Appendix
E. If you are using HRTF to calculate IACC, you will need to create a filtered set of HRTF’s for this
purpose. Please read section 6.5 about IACC.

Headphones
The headphones used for auralisations will influence the experience, because any pair of
headphones will change the frequency response of auralisation. Therefore, ODEON can minimize
the influence of the headphones used by applying an inverse filter. Either choose one of the given
headphone models from the list or add your own. There is a big difference on different headphones,
so it is important to choose the correct ones if possible. Remember that if you want to show the
auralisations to a client, the auralisations should be made for the type of headphones the client is
using.
For many users it is not possible to either measure the headphones used or find a matching one
included with ODEON. If you do not have a matching headphone filter – then you should instead
use the diffuse field filter matching the set of HRTF’s selected. Odeon automatically creates generic
diffuse field filtered headphone filters for each set of HRTF’s. If e.g. using the Subject_21Res5deg_...
HRTF then the matching headphone filter has the name Subject_021Res5deg_diffuse.

90
The headphone filter design is explained in more detail in Appendix D, as well as how to add the
frequency response of your own headphones to ODEON.

Besides these, there are other options in the Auralisation setup and these are explained in the help file.
Press F1 when you are in the Auralisation setup, and a help box will appear.

Remember: Always press F1 when you need to learn more about specific functions in ODEON. The
context sensitive will open at the page related to the active window. Often, the ODEON help file
complements the manual and offers extra information. To learn more about Surround playback and
how to define a speaker rig for surround playback – which is not covered here - click the Define speaker
rig button then press F1.

Ambisonics settings
This is the lower half of the auralisation setup window and includes parameters for reproduction
through the ambisonics technique. This makes it possible to listen to auralisations not only through
headphones but also through loudspeakers.
ODEON can produce either a 3D B-format output, which can be decoded into any loudspeaker
system in an external program or a 2D surround output (decoded already) which can be played
directly to a horizontal loudspeaker system. Remember to hit the F1 key to receive valuable
information from the ODEON help.

Filter settings
The section controls parameters for the filters used reduce calculation time, as with Binaural
auralisation (see previous section). As with HRTF, for most users the ambisonic filters are not very
important to learn about in detail. You should in principle keep the default settings. However, you
can read more the details and how to change them in the Appendix E.

Encoding
Specifies the order of the ambisonic encoding. The higher the order the more precise the encoding
will be. Three orders are available:

1st Order

This provides 4-channel wave files, WXYZ, with W being the omni.

2nd Order

This provides 9-channel wave files, ordered according to the Ambisonics Component Ordering,
ACN (corresponding to Furse-Malham): WYZXVTRSU (used to be ordered WXYZRSTUV in
Odeon 14 and earlier versions)

3rd Order

This provides 16-channel wave files, ordered according to the Ambisonics Component Ordering,
ACN (corresponding to Furse-Malham): WYZXVTRSUQOMKLNP.

Note: Currently it is advisable to use 1st Order ambisonics for your ODEON loudspeaker auralisations.
Higher orders should in principle give more detailed reproduction, but the implementation of the Furse-
Malham algorithm results to a somewhat blurred sound field.

91
B-format settings
Use these settings to create a B-format output, which can be decoded outside ODEON to any 3D
or 2D loudspeaker system.
Check the box Create impulse response (*.BFormat) to create impulse responses corresponding to the 3D
ambisonic channels, defined in the Encoding. These impulse responses can then be convolved with
anechoic files in the Job List to provide BFormat convolutions (auralisation files). The Bformat
convolutions are multi-channel files (4,6 or 16, depending on the chosen order) and have to be
decoded to a loudspeaker system in an external software application.

2D surround settings
Use these settings to create a surround output – already decoded, which can be played directly to
any 2D surround system that consists of a number of speakers and a receiver (eg. home cinema
audio systems). The tool does not provide 3D decoding.
Check the box Create impulse response (*.SurRoundnn) to create impulse responses that correspond to
each speaker in the defined speaker rig (see just below). These impulse responses can then be
convolved with anechoic files in the Job List to provide 2D surround convolutions (auralisation files).
The Surround convolutions are multi-channel files that include one channel for each loudspeaker
system, and can be played by any media player such as QuickTime, Windows Media Player and so
on.

Compensate speaker delays

If all loudspeakers are not at the same distance from the listeners position (sweet spot), ODEON
will delay some of the speaker feeds as appropriate. Enabling the function makes the quality better
in the sweet spot - at the same time it might become worse outside the sweet spot.

Parametrization (2D)

A system of 3 equations with 3 unknown is solved for 1st order 2D ambisonics, while a system of
5 equations with 5 unknowns is solved for 2nd order 2D Ambisonics. The results are 3 or 5
constants; one constant for each part of the Ambisonics equation. The benefit is that the decoder is
then capable of recreating plane waves correctly even if a regular speaker array is NOT applied.
By a regular array is meant a speaker rig with equal angles between all speakers.

Remember: Make sure to check the Create impulse response (.SurRoundnn) in the 2D surround settings panel
to make a surround decoding inside ODEON. If you want to export B-format auralisations in order to
decode into a loudspeaker setup outside ODEON (mainly 3D), make sure to check the Create Impulse
response in the B-format settings panel.

The following buttons are available in terms of defining a speaker rig:

Define speaker rig

Here you can define the actual setup of your surround system. Some templates are available, with
the most common being perhaps the 5.1 and 7.1. The increment .1 indicates the use of subwoofer,
which comes as an additional (omnidirectional) channel in the file (Check/Uncheck the box Use
subwoofer). The pre-defined templates are either circular or rectangular, but the coordinates can be
adjusted on the right side to create a fully customised setup.

92
Use program default settings
Use this to restore the settings like the Overall recording level, the Compensate speaker delays
etc.

Make surround settings program defaults

Use this to assign the custom settings for Overall recording level etc. as default.

Save speaker rig and make archive file

Use this to save a custom speaker rig, created in the Define speaker rig.

Get speaker rig from file

Use this to load a speaker rig previously saved with the previous function.

The Jobs, sources and receivers

Source and Receiver options
Each auralisation can only be made for one single receiver, as one can only be listen at one given
position at a given time. The direction of the receiver also important, as it will be the orientation of
the receiver (being the simulated dummy head in auralisation).

• If you want to make an auralisation with multiple sound sources from different directions
with different sound files, make a job for each sound source or use the “Multi source/signal
auralisation expert”.

• If you want the same sound file from multiple sound sources, you can just make one job
including all the sound sources.

• If you want different sound files from same position, you can also just make one job, and
use it for multiple convolutions later.

Single Point Response

Some auralisations made in ODEON are based on the Single point response receiver jobs in the Job list. If
using these, you have to make jobs with one or more active sound sources and a single receiver.

93
A note on directivity patterns for natural point sources
It is common to make auralisations with natural point sources. With natural sources we refer to
sources such as human voice, an acoustical instrument or similar, where a recorded signal for
auralisation may be associated with a certain directivity pattern. In order to select the right source
for auralisation purposes, please read Chapter 3.

Sound cards
A sound card is required in order to play back the auralisation results and may also be useful if
you wish to transfer (anechoic) signals to the hard disk. As a minimum, the sound card should be
capable of handling signals in stereo, in a 16-bit resolution at a sampling frequency of 44100 Hz.
Also consider whether the card is immune to electromagnetic noise, which is always present in a
PC and whether the analogue output for headphones is satisfactory. (For surround auralisation,
obviously a multi-channel surround soundcard is needed along with the necessary loudspeakers
and amplifiers).

7.4 Making auralisations

There are different ways of making auralisations. The way to go will depend on what your needs
are. The needs could be:

• A fast auralisation tool to listen to different jobs. E.g. in a design phase where different things
are tried out in the model. This can be done with Streaming Convolution.

• An exported .wav file to show/send to costumers or publish on a web-site. This needs to be done
via an Offline Convolution.

• A single audio file with with many different signals being played from different sound sources
at the same time. This kind of multi source auralisation requieres to be created as an Offline
Convolution.

• Multiple audio files played simultaneously, from different sound sources and maybe a delay
on some of the sources. This needs to be an offline convolution, and for most cases it will be
easier to use the Multi Source/Signal Auralisations Expert.

• Making eg. a canteen or cafeteria where there is a need for many sound sources, placed all over
the space, or a full orchestra playing a classical piece. You would probably both need the Multi
Source/Signal Auralisations Expert to make the setup, and the Audio Effects tool to turn a few
audio files into multiple, like a few voices into a crowd, or a few instruments into an orchestra.

• Play an impulse response for a job in the Job List using Play Single Point BRIR. You can also
save it and reload it, if you need the measurement options.

94
Play BRIR – listen to a Dirac signal
Once a Single point response is calculated, it is possible to play the BRIR, clicking the Play Single Point BRIR
button. The BRIR may give a first clue as to how the room sounds and it also allows some
evaluation of the quality of the calculated point response, e.g. whether to use a higher Number of
late rays in the Room setup.

Although the BRIR may just sound like a click or pop, it may work quite realistically when
convolved with a signal less transient than an ideal Dirac delta function. To get a more realistic
presentation of a BRIR as it would sound in the real world you might want to convolve it with the
Clapping signal file, an anechoic recording of clapping hands, which is a less transient signal than an
ideal impulse.

Streaming auralisations – listen to different audio files

For fast evaluation of the acoustics, use the streaming convolution. In the Joblist, on the Jobs tab,
you can choose Streaming convolution in the menu on the right-hand side. The streaming will be
for the selected job on the Joblist. A menu will open, where you can listen to streaming
convolutions.
The main advantage is that you can easily listen to different audio files while the room acoustics of
your model is applied.

Under Input signal, you choose the audio file you would like. If you press Listen to input signal, you
hear the anechoic signal, otherwise you listen to the convolved signal for the selected job. Under
Channel setting you can choose to use a stereo source. This is explained in the help file (press F1 when
the Streaming convolution menu is open), but for most cases it is recommended to stay with the default
Mono source.

Offline Auralisations – Most Options and Export .wav files

For the best flexibility, mixing different convolutions together, making individual adjustment of
each channel (delay and level), and storing auralisation results in .wav file, you should use offline
auralisations. Audio files produced in the offline auralisation may be used online, in Power Point
presentations, etc.

Making an offline auralisation

Since ODEON 15 there are a couple of ways to make an offline auralisation. Either you set
everything up yourself, you use the Auralisation Expert that will generate everything for you, or
a mix of both.

• Use manual setup if you want the auralisation for a specific job in the Job List, eg. to be able to
compare auralisation with parameters.

• Use the Auralisation Expert if you want to choose a receiver position, multiple source
positions and multiple audio files, and let the expert make the rest.

Both the normal setup and the Auralisation Expert are described in this chapter.

The auralisation Tab Sheet overview

If you have check-marked Create binaural impulse response file (or Create 2D Surround Impulse Response) in the
Auralisation setup, you will have an extra tab in the right-hand side of the Job List called Auralisations.
In this tab you will be able to make auralisations from the jobs you have in the Job List.

95
The auralisation tab consist of a Convolver menu, a Mixer menu and then a range of options on the
right-side tool bar, including the Auralisation Expert, added in ODEON 15.

In the Convolver (left part), Jobs (Impulse responses) and audio files are paired in different convolutions. In the Mixer
(right part), convolutions from the left part can be mixed together.

Warning: The auralisation toggle button is disabled if Create binaural impulse response file (.Jnn) or Create
impulse response (.SurRoundnn) is disabled in the Auralisation Setup (SHIFT+CTRL+A).

Convolver
If you are working with auralisations for headphones, you have the Convolve Binaural RIR with signal file
in the left side. This is where the single auralisations is made, if not using the Auralisation Expert.
Here, mono signals are convolved (paired) with Binaural Room Impulse Responses (BRIR's, which
have been calculated as part of the Single Point responses). This process may be compared to a
binaural recording of a mono signal, played through simulated sources in the room.

The menu consists of:

• Signal Sub Path are the folders of sound files. If blank, you have a range of different sound
available, but you can also choose different classic pieces played by different instruments if you
are making a full orchestra.
• Signal file is the file is file containing the sound you use for auralisations. All signals that come
with ODEON are anechoic recordings. If you add new sounds, make sure it is (or close to)
anechoic.
• Time is the duration length of the recording.
• Channel displays the channel of the audio file being used. If it is a stereo audio file then you can
choose whether to use the channel 1 (left channel), channel 2 (right channel) or the default which
is an average of the two. Audio files with more than two channels can also be handled.
• Calib. When the Calib. check mark is checked (see Figure 5), ODEON will 'calibrate' the level of
the input to have a fixed A-weighted level; so two different recordings, e.g. voice and noise, will
have a comparable loudness (they are adjusted to have the same A-weighted level before they
are convolved with the impulse response).
• Job no. is the job number of the impulse response (Single Point Response) in the Job List used for the
convolution. The Single Point Response defines the receiver position and orientation and which
sources are active.
• Receiver and Job descr. is the information about the job, and the corresponding receiver, chosen as
taken from the Job List.
• Rcd. Lev. is the recording level. By default it is set to 0 dB but you can change it if the convolution
is too loud or weak when calculated.
• Max. out displays the peak of the convolution calculated. It corresponds to peak level that can be
displayed by a recording device used in the real world. The Max. out level must be close to 0 dB,
for an optimal dynamic range, but never above 0 dB, in order to avoid digital clipping
(distortion – ultimately white noise if the overload is substantial). If the calculated convolution
is too loud or weak, use the Rcd. Lev. to change, then recalculate.

96
Mixer
The mixer is located in the right side of the auralisation tab. The mixer allows convolved results to
be combined into one wave file allowing multi-channel simulations, e.g. stereo setups, singer
versus orchestra etc.
The mixer consists of two lists. The first is the list of all mixes, Mix convolved wave results into one wave
file. For each mix, there is a list of convolutions in contains - Convolutions in mix no. <xx>.

Remember: Only convolutions that use the same receiver can be included in one mix. This
corresponds well with reality where our head remains momentarily fixed at a position when hearing
multiple sounds. Even when listening to a full orchestra, our ears look at one instrument at a time.

The two lists consist of:

• Mix no. is the number of each mix which can help locate the mix later, when exported as wav
files, as the number will be part of the .wav file name.
• Enabled Turn on/off a mix. Most important when using the Run All Jobs option.
• Rec. is the receiver No., corresponding to the selected convolutions in the mix. Only convolutions
with the same receiver can be included in the mix.
• Description is a description of the mix. Recommended if having many mixes.
• Convolved files are the numbers of the convolution included in the selected mix.
• Max out is the peak level in dB of the mix. This needs to be below 0 in order to avoid digital
clipping but is good to keep close to 0 for better sound quality.

Convolutions in mix no. <xx> panel:

• No. Consecutive numbering of the added convolutions.
• Conv. no. is the number of each convolution you add to the mix. You can add up to 300
convolutions which may use different sound files and sources, as long as they have the same
receiver position and orientation.
• Mix. Lev This is your option to set the different levels if making a mix.
• Delay in sec. You can delay some of the convolutions. Used mainly for a delay effect e.g. in
churches or large rooms or outside where the sounds needs to be delayed between different
speakers in order to avoid echo/delay effects. Can also be used to make the same convolutions
sound like several, to use in canteen or office cases.
• Job descr. displays the description of the job/impulse response used for convolution.
• Signal file displays which signal was used for the convolution.

Multi Source/Signal Auralisation Expert

Setting up jobs (impulse responses), convolutions and mixes e.g. for a crowd in a canteen or an
orchestra in a concert hall using the tables in the Job List can be tedious and extremely time
consuming. The Auralisation expert makes that process much faster. The Auralisation Expert can be found
in the right-hand side toolbar of the Job List, but you need first to select (with a mouse click) the mix
you want to edit in the middle table on the Auralisations tabsheet. So, select a mix number and then
press the Multi Source/Signal Auralisation Expert tool button.

97
The Auralisation Expert is a really easy way to convolve multiple audio files with multiple sources,
making auralisations at a specific receiver position. This tool works very well with the tool Audio
Effects that can create multiple different audio files from a single audio file, to make a crowd in a
canteen or orchestra in a concert hall, among other scenarios. The Audio FX tool (Audio effects)
which is explained later in this chapter.

The options in the Auralisation Expert are:

• Mix description is an optional description describing the auralisation mix, e.g.: Crowd for canteen
or Orchestra playing Mozart at conductors’ position.

• is the listener position in the room. All convolutions included in the mix will be
Receiver
simulated with this position.

• defines the direction of the receiver influences the results, as the auralisation is
Receiver looking at
either binaural or surround. You can choose between x-axis direction, or point it toward a
source position. If you have a specific place you want the receiver to look at, where there is no
source, you can make a dummy source in the source receiver list.

• The source number is on the left side, referring to the source number in the Source
The list of sources
receiver list.
You can choose a number of sources with ctrl + mouse click, and when you import
files (if the same number of files is imported), these files will be assigned to these source
numbers. All sources that have been defined are included in the list, but indeed all the sources
does not need to be included in the mix – only the sources to which a sound signal file is
assigned are included in that particular mix. Each source will be assigned its own signal and
normally a signal file is only used for one particular source (e.g. the 12 th violin signal is only
played by one violin/musician).

• Select multiple signal filesin the upper-right corner of the Auralisation Expert allows assigning multiple
signal files to multiple sources. First select sources in the list using Ctrl+mouse click then click
the Select multiple signal files icon and select the audio files in a similar manner in the file dialog that
appears. When closing the file dialog, the files will be added to the selected sources in the list.
If you fail to select a matching number of sources and signals Odeon will do its best to match
them anyway – and usually this works out OK.

When you have closed the Auralisation Expert, Single point responses, convolutions and mix has been
setup but not calculated yet.

98
 Run the calculations
When a mix has been setup in the Auralisations tabsheet, you can calculate it by selecting it in the
Mixer and pressing Run Single Job, or pressing Run All Jobs. The appropriate rows in the mixer will turn
green when results have been calculated.

Play or export wave file

Once the calculations have been carried out, click the Play wave result button and listen to the result
through the headphones. If you have selected (left click mouse) the Signal Sub Path or the Signal file
column in the Convolve BRIR and Signal file table, the (anechoic) input file is played. If you have selected
Receiver or Job descr. it is the convolved file that will be played. If you have selected the mix, anywhere
in the Mix convolved results into one wave file, it will be the mix that is played.

If you want to export the wav file, there are several ways to do it. You can save it from the Wave
player when you are in the Play wave result player window Another way is to simply drag-and-drop
the wave file from the appropriate cell in the Job/Auralisation list into the desired destination e.g.
a folder in Windows Explorer your Desktop or directly into Microsoft Power Point. If you need to exchange
many wave result files, then you can find the wave file in Windows Explorer, open the appropriate
room folder using the File|Explore current room folder (hint, type RoomName*.*wav in the search field of
Windows Explorer to locate the relevant wave result files – RoomName being the actual name of your
current room)

Remember: The play wave function is sensitive to the cell and row chosen in the table. When the input
signal cell is selected the play function reproduce the input signal without convolution. You have to
select one of the cells that have to do with the Job (like Job no., Receiver, Job descr.) in order to listen to
the convolution result.

In summary, a quick guide on how to make auralisations with the auralisation expert:
1. In Auralisation Setup, check-mark Create binaural impulse response file.
2. In the Job list, open the Auralisations tab at the right side.
3. Select mix and open the Auralisation Expert.
4. Choose a receiver and direction.
5. Select the source numbers you want to add audio file to.
6. Open audio Select multiple signal files and choose audio files.
7. Close the Auralisation Expert.
8. Calculate the mix.
9. Listen and/or export a .wav file.

Ambisonics Auralisations
When you have enabled either B-format or Surround
impulse response creation in the Auralisation setup
(Sec. 7.3) you have to make sure to click the toggle
Binaural/Surround/B-format mode button on the right
toolbar to switch between the different modes.

B-format mode
The B-format mode only produces B-format files to be decoded in an external decoder, therefore it
is not possible to play them through ODEON.

99
Instead, in this mode you can drag and drop the B-format files to any folder you prefer for further
decoding processing.

Surround mode
In the surround mode you can simply play any convolution through ODEON directly by selecting
the corresponding convolution and pressing the player button . You can even export any
convolution, as with any binaural Auralisation, simply by drag-and-dropping into the desired
folder.

Remember: The B-format mode does not play any file. You can only drag and drop B-format files to be
processed in an external decoder. Instead, use the Surround mode to play 2D ambisonic convolutions
through the built-in ODEON player.

7.5 Audio Effects – Multiplying the Audio Files

A common problem experienced when
making auralisations is the limited number of
appropriate anechoic audio files available.
For realistic multi-source auralisations (e.g. a
cafeteria, an office or a large string section),
many recordings are needed (eg. speech or
violin parts).
Although ODEON comes with an extensive
list of anechoic audio files, they often need to
be manipulated, so that their variations can
be assigned to different convolutions of a
multi-source auralisation. Cloning and
overlaying the exact same audio instead
would not provide a realistic effect and an
impression of several uncorrelated sources.
This workflow of creating variations and
individual convolutions is explained in
(Christensen, et al., 2021).

An Audio FX (Audio effects) interface has been created since ODEON version 15, which facilitates
the creation of many variations out of single audio files. Six effects are currently available: Change
Duration, Phase Vocoder, Chorus, Spectrumizer, Gain and Amplitude Modulation.
Traditional audio effects tools are aimed at making a single fine-tuned variation of an input file,
based on some user-defined parameters. However, the purpose of the ODEON Audio FX tool is to
make a population of output files automatically, with random variations of the associated

100
parameters. Like in any large population, it is expected to have outliers. So if for example one out
of many voices in a population ends up with a very low tempo or a very low pitch, then this is not
considered an error – but an expected behaviour for the population.

Note: All the effects come with some pre-defined parameters, that have been finetuned by the ODEON
team, after long and repeating listening sessions. However, every multi-source auralisation scenario is
different, therefore it is worth trying different settings.

It is possible and often desirable to use multiple effects on the same input – though you will
probably not be using all effects at the same time. On string instruments it is useful to apply the
Chorus as well as the Spectrumizer effects to simulate different players playing on different instruments.
For voice samples to be used for crowd simulations, it is often desirable to use the Change duration
and Phase Vocoder effects to created outputs that are incoherent and with a desired duration and voices
of different pitch and tempo.

When using the files generated in the Audio FX’s tool for multi-source auralisations, the easiest way
to do that is to use the Auralisation Expert , and just choose all the generated files at once.

How to Use the Audio Effects Tool

The Audio Effects tool is located at Tools>Audio FX’s. The interface looks like in the following picture:

1. Open one or multiple audio files (by SHIFT+LMB or CTRL+LMB). You can open multiple audio files,
and for each of them the chosen number of new output files will be generated.
2. The waveform of the first of the selected files is displayed. If multiple files have been selected,
only the waveform of the first is displayed. However, the rest are still kept in memory for further
processing.
3. Press the + tab sheet to choose an audio effect.
4. Change the settings for the chosen effect (press the F1 key to access the ODEON’s help for details
on the different parameters).

101
5. Press the +tab again to add more effects in the chain and change the settings if necessary.
6. Choose Number of output files for each audio file. This many different audio output files will be
generated for each input file. For each generated file, the settings for the effects will be varied
randomly within the chosen limits.
7. Press Play audio file to listen to the effects applied on a single audio file. If loading another input
file then press the button again to listen to the new sample with the effects applied. If you change
the effect settings, this will not take effect until you press the Play audio file button again.
8. Press Save to generate and save the Number of output files. Choose output file name(s), press Save
again and then please be patient. It will take a (little) while for the audio files to be generated.

List of Audio Effects in ODEON

The Audio FX tool currently includes six effects, which will be described below (Christensen, et al.,
2021):

Change Duration
The Change Duration effect changes the overall duration of an input file, either by adding repetitions
and silences between them (longer duration) or by cropping it (shorter duration). Therefore, the
effect’s goal is not to alter the pitch or speed of the file.
This effect can be useful in creating scenes and soundscapes containing multiple speech signals,
music, noise events etc., which will typically require that the input signals to the multi-source
auralisation have the same length.
The user determines the final duration of the file, the length of silence (or a range of random
silences) between repetitions and the length of fade in/out for the whole output signal.
The effect is mostly useful for creating not only one, but a population of output files with a longer
final duration and random variations of silence between them. Such a procedure is done according
to the following algorithm:
1) A random location is determined between the start and end of the input signal. This location is
not controlled by the user.
2) A random period of silence is added at the end of the file, according to the maximum limit given
by the user (the default value is 3 s). This is useful especially when working with speech. A copy
of the input signal will follow the silence period. If the user specifies a negative
silence period, then the repetitions of the signal will overlap, without a silence. To accomplish this,
ODEON applies cross fading between the repetitions. A negative period of silence is useful when
attempting to extend the duration of a stochastic signal, such as noise.
3) The first two steps are repeated until the length of the output signal is exceeded and the output
signal is truncated at the desired length.
4) Fade in/out is applied at the beginning and at the end of the output signal, in order to avoid
clicking sounds and give the desired duration to the output file. The following figure illustrates the
process. Since other effects may change the length of the signal, the Change Duration effect is
always added as the last effect in the chain, to ensure that the output will end up having the
specified duration. An exception to this rule is when the Amplitude modulation effect is used in
the chain. Two reasons justify this exception: 1) It might be desirable to use this effect after the
Change Duration effect to vary the gain across the whole final
length. 2) The Amplitude modulation is applied directly in the time domain signal and does not
alter the original duration.

102
 Principle of the Change Duration effect.

Phase Vocoder
A common way to change the pitch of a signal is by resampling it in the time-domain. For example,
when a signal is downsampled (fewer samples per second), then the pitch rises, but at the same
time the duration becomes shorter (higher playback speed). On the other hand, if the signal is
upsampled (more samples per second), then pitch becomes lower, but duration becomes longer.
However, it is often desirable to change the pitch and speed independently. The Phase Vocoder
belongs to the time frequency processing audio effects and among others it offers that particular
function. It can be used for example to make a speech signal slower, without changing its pitch.
The effect transforms the time-domain signal into a time frequency-domain representation, where
its Magnitude and Phase can be modified. The modified frequency-domain representation is then
converted back to the time-domain output signal.

Chorus
The Chorus effect is the result of multiple signals of similar pitch and timing played together and
perceived as one. Unlike the classic chorus effect - where one output file is produced, the chorus
effect in ODEON produces a separate audio file for each signal in the group. To achieve a more
realistic variation of pitch and timing, the implementation of the Chorus effect in ODEON is based on
fractional delay lines (Zolzer, 2011) that have random variations as a function of time.
Typical chorus effects include the section of strings in an orchestra or a choir of voices within the
same vocal range.
Although musicians in a string section as well as the singers in a choir try to perform the exact
same musical piece, there are always slight variations in pitch and time of musical notes. This is
because the timing and performance of the individual musician or singer are different from other
musicians/singers, while the instruments used can differ significantly from each other.
We assume that the time each musician starts playing/singing (onset time) follows a normal
distribution, with a standard deviation σ of 40 ms. According to that, assuming that the actual onset
time (the mean of the normal distribution) is 0.00 ms, approximately 68.2 % of the onset times will
occur within 40 ms (1x σ), 27.2 % within 80 ms (2x σ) and 4.2 % within 120 ms (3x σ). The chorus

103
effect in ODEON is implemented according to the following algorithm, utilizing a fractional delay
line:
1) A vector of random delays as a function of time is constructed with the same length as the
input signal. The vector is constructed in two steps: a vector of control points is made with the
points placed at regular intervals specified by the frequency parameter of the effect. Each
control point is assigned a random delay which varies around a central delay between _ a
depth parameter, given in percent. A final delay vector is constructed by interpolation
between the control points.

2) The delay times in ms are converted to non-integer sampling intervals, in order to keep
continuity in the delay vector. For example, if the sampling frequency fs is 44100 Hz, a delay
of 100.309 ms is converted to 4423.627 samples.
3) A sampling interval delay line x with length equal to the maximum number of samples in the
delay vector is constructed.
4) Each sample of the input signal is moved by the exact number of samples of the corresponding
cell in the delay line. Since the movement is done by a non-integer number of samples,
fractional delay is be utilized (Christensen, et al., 2021).

Spectrumizer
The Spectrumizer makes random changes to the spectrum of the input, similar to an Equalization
(EQ) filter. The main input parameter is the Spectral Random fluctuations in percent which is by
default 50 % meaning the level may be decreased up to 6 dB at some frequencies and increased up
to 3 dB at some frequencies.
By adjusting the levels in percentage, rather than in a dB scale, it is ensured that the average level
of an output signal is the same as that of the input signal which is important for a controlled design
of soundscapes. The spectral changes are made at a number of frequency points specified by the
user. The points are distributed logarithmically leading to the same degree of randomness per
octave. The maximum number of random points per octave sets a lower limit to the lowest
frequency that will be modified.
The effect can be used to alter the timbre (colour) of musical instruments. For example, it can help
creating variations of the individual string instruments in a string orchestra (violins, violas etc.). It
can be used in combination with the Chorus effect to create a full string section out of one input file.

104
Gain
The Gain effect is very simple. It changes the gain by a fixed number of dB.

Amplitude Modulation
The Amplitude Modulation effect changes the amplitude of the signal by a random envelope that is
defined between a minimum and a maximum change in dB. The following figure shows a typical
random variation of the gain applied to the signal throughout its duration (10 sec). We can see that
the gain varies approximately between 0.4 and 1.6, which corresponds to a maximum of ± 60 %
change of the original power. A series of random-gain change control points is constructed at a
fixed period of 0.5 s, which corresponds to the random period that the algorithm selected between
the specified range of 0.1 s and 0.6 s.

7.6 Examples of Auralisation

Example 1: Single Channel Simulation
Try to create a one-
channel simulation of a
person speaking from a
source position.
1. Open the room Elmia
RoundRobin2 detailed.par,
which is stored in the
ODEON Rooms folder
(see Chapter 2.1).
2. In the Auralisations tab (far right-hand side of the Job list), in the Convolver side, select the Signal file
field in the Conv. no.1 row and select the Voice Sabine Short file (this is an anechoic recording of a
voice stored in a Windows Wave file format - residing in the directory set in the Options|Program
setup|Paths|Wave folder field).
3. To play the chosen input file after selecting it, click anywhere else in this row (this finalizes the
selection) then press the Alt+S shortcut (or the Play wave button ) to listen to the file.

105
4. Choose a Job from the Job no. column to convolve an impulse response with the selected signal.
Exit the Job no. cell by clicking anywhere else in the same row. Information about the job will
then appear.
5. Press the Run Single Job button . A calculation of the point response between the source(s) and
the receiver in the selected Job is made if not already calculated. Afterwards the convolution
between the response, and the selected anechoic signal in the list is performed.

If the convolution line

turns red at the end of
the process, then output
level is too high (above 0
dB) and therefore digital
clipping will occur. This
has to be addressed;
otherwise, you will
experience unpleasant
distortion when listening to the auralisation. You can reduce the recording level in the Rcd. Lev.
column by the same amount as the exceeding dB. In the example the level exceeds by 62.36 dB.
Setting the recording level down to -63 dB will lead to an output level just below 0 dB. This process
is very practical for single files, as the absolute level of the playback can be further adjusted by the
windows volume control. However, when mixing different convolutions, you might need to
preserve their individual levels. For example, if you make an auralisation of an orchestra, where
all instruments have been recorded separately, their individual dynamics (levels) should remain
unchanged. To avoid clipping you should reduce the Overall Recoding Level in the Auralisation setup. Once
this is addressed, you may listen to the result by selecting the convolution’s Conv. No. in Convolutions
in mix no. 1 in the mixer side, clicking elsewhere in the row, then pressing the Alt+S shortcut (or the

Play wave button ).

Remember: When the output level is higher than 0 dB undesired clipping will occurs. In this case you have
to reduce the recording level (Rcd. lev column) or the Overall Recording Level in the Auralisation Setup.

Example 2: Auralisation of Canteen Area Using Audio FX

With the Audio FX tools and the Auralisation Expert, making an auralisations of a crowd is much easier.
In this example, a realistic-sounding canteen area will be made from only a few audio files.
The example is a simplified model of the canteen we use at ODEON. It has 6 tables with 12 seats at
each. In this auralisation we will imagine a full canteen with half of the people speaking at the same
time, so a mix of 36 convolutions. We are only going to use 3 anechoic audio files. You can even
make a decent auralisation with only one anechoic input file on the other hand using more than
three files could probably an even better auralisation.

1. Press File > Open Room and find the CantineExample/CantineExample_Ready.par model
2. Go to Source Receiver List in the top menu (or shortcut Shift+Ctrl+S)

106
3. Select the source People facing north and press the Tabulate selected source or receiver tool in the right-
side menu. Set First dimension -X to 1.4 m, Nx to 3, Second dimension -Y to 2.2, Ny to 6. This way
it will make 3 sound sources placing north for all tables. Do the same thing for the People facing
south.
4. Open Tools>Audio FX
5. Open audio files Agora, “Den Musikalske Erimitage” and “Voice Sabine” (use Ctrl+left click). They are
all recordings of the same person, so they do sound very similar. Different voices would give
a better effect. To select multiple file, hold down the Ctrl key while clicking the files.
6. Add the two effects; Change Duration and Phase Vocoder. Set the final duration to 1 minutte (and 0
seconds) in the Change duration effect and leave the rest of the default settings.
7. Set the number of output files to 12.
8. Press Save.
9. Open the JobList (shortcut Ctrl+Shift+J).
10. Open the Auralisations tab at the far right-hand side of the window.
11. Select Mix. No. 1 and open the Multi Source/Signal Auralisation Expert .
12. Change Receiver to the receiver Listening position
13. Rename the Mix description, for example, to Listening Position” (or any name you desire)
14. Use the Select multiple audio files icon in the upper-right corner of the Multi source/signal auralisation
expert, to assign an audio signal for all sources. Only use the signal files which you just created
in the Audio FX, so the new versions of Agora, Den Musikalske Erimitage and Voice Sabine (having the #
sign and file number appended to the original file name).
15. Close the the Multi source/signal auralisation expert
16. Select Mix nr. 1 and press the Run single job icon (or shortcut Alt+R) in the right toolbar.
17. Press the Mix. No. 1 (which should be green by now) and press the Play wave icon (or shortcut
Alt+S) in the right toolbar.
18. Listen to the auralisation, and save it if you want. Now you can try to change the materials of
the room, or something else, and hear how the auralisations will change. You have to
recalculate the auralisations after changing materials, but you do not have to set it all up again.

Example 3: Extended Auralisation with Complete Orchestras

Since ODEON version 10 a number of auralisation setups are available, where entire orchestras
have been modelled for auralisation of a concert hall and in an opera. These are available in the
WavePack 1 Install which you can download at: https://odeon.dk/downloads/anechoic-recordings/.
The installation includes pieces of Beethoven (21 channels), Bruckner (47 channels), Mahler (39
channels) and Mozart (11 channels). In order to listen to these auralisations:
• Download and install the WavePack. Remember to generate the required additional files by
loading the batch file using the AudioFX tool. The instructions are in the file
“OdeonWavePackLicense_I.txt” in the created WaveSignals\Orchestra2019 folder.
• Open one of the rooms in the Orchestra2019 folder (e.g.
C:\Odeon…\Rooms\Orchestra2019\ComptonVerneyOpera-Mozart.par)

• Open the JobList (shortcut Ctrl+Shift+J) and click on the Auralisation tab at the far right-hand
side of the window.
• Calculate all jobs in the JobList (shortcut Alt+A).

107
• Once the calculations have finished in the Auralisation display, you can play the various tracks
using the Play wave icon (shortcut Alt+S).

The auralisation samples include many setups that inform ODEON how jobs and wave files should
be combined into a few final mixes. If you would like to listen to auralisations in other receiver
positions than the predefined ones, the fastest way to do this is:
• Enter the Source receiver list (shortcut Shift+Ctrl+S) and define the new receiver (shortcut
Ctrl+R).

• Open the JobList (shortcut Ctrl+Shift+J).

• In the Auralisation display, duplicate the mix by selecting Mix. no. 1, copying (Ctrl+C), selecting Mix.
no 2, and pasting (Ctrl+V).
• Open the Multi source/signal auralisation expert and select the new receiver. When you close the
window, ODEON will automatically create new jobs, convolutions and a mix as needed.
• Calculate all jobs (shortcut Alt+A).

The anechoic recordings were created by Helsinki University of Technology. For use of the
anechoic orchestra recordings, the conditions as specified in the software license agreement must
be observed, see page 3.

Example 4: Auralisation of Speakers Playing Music

The next example will demonstrate how to set up a ‘classic’ stereo setup with a receiver position
and two loudspeaker positions. This example is very simple and does not come with a room
prepared for it, so choose any room that you like. Either one you create yourself or one of the rooms
that comes with ODEON.
• To run this example, you need to have a stereo recording stored on your hard disk as a Windows
wave file, at a sampling rate of 44100 Hz. You should be able to use most recorded music files.
• This file, which does not need to be an anechoic recording for this demonstration, should be
placed in the directory set in the C:\Odeon...\WaveSinals\ directory.
• Open a room. Choose any room you want to place a speaker system in.
• Enter the Source receiver list (shortcut Shift+Ctrl+S).
• Make a New point source (shortcut Ctrl+P) at the position of one of the speakers. Assign a
Directivity pattern to the loudspeaker or leave the speaker with the default of Omni.SO8 if so desired.

Note that you’ll have to change the Sub directory of the Directivity pattern to access a preinstalled
directivity pattern of a loudspeaker.
• Make a similar source at the other speaker position (you may use the copy icon (shortcut
Alt+C)and changes coordinate(s) of the new source.
• Make a receiver position for the auralisation (shortcut Ctrl+R).
• Go to the Job list (shortcut Ctrl+Shift+J ) and into the Auralisations tab.
• Select a Mix No. and click the Multi Source/Signal Auralisation Expert .
• Choose the same signal, the stereo music file, for both sources made.
• In the Channel option, choose channel 1 (left) for the left source and channel 2 (right) for the right
source.
• Close the setup, calculate the mix and play the auralisations or export it.

108
7.7 Sound Pressure Levels for Auralisations
Headphones
The binaural auralisation results created in ODEON are by default binaural signals which should
be presented over headphones. The purpose is to reproduce the same sound pressure at the
entrance of the ear canals (and at the eardrums for that matter) as is obtained in the real room (if
the room exists).

Adjusting levels
Sound Pressure Level is one of the most important room acoustical parameters, so it is important
that levels, at which auralisation samples are presented, are realistic. If playing a simulation of
voice at an unrealistic high level, the speech intelligibility may be over rated. It does not help that
Clarity or Speech Transmission Index is satisfactory if the Sound Pressure Level is too low. If
playback levels are too high, echo problems may be exaggerated, because echoes that would be
below audible threshold (or at least at a very low level) are made audible. Another very common
effect of presenting auralisations at too high levels is that the room sounds too reverberant. This is
because the full decay may be audib–e - even some 90 dB of decay may be audible where the
definition of RT assumes that at its best only 60 dB of decay is heard.

The levels presented in auralisation samples created by ODEON are influenced by:
• The HRTF’s.
• Level in input signal file, e.g. the RMS value or LeqA.
• Calculated Sound Pressure Level, which is based on geometry, sources, receiver positions,
materials, etc.
• Overall recording level in the Auralisation setup.
• Recording. Level (Rcd. lev) in the Auralisation display of the JobList - if off-line convolution is
used.
• Mixer levels (Mix. Lev. in the JobList) if offline convolution is used.
• Gain in the Streaming convolution dialog if the real-time convolution option is used.
• Output gain of the sound card, the volume setting.
• Sensitivity of the headphones.
• Coupling between headphone and the subject’s ears.

Maximised playback levels for maximum dynamic range

If you only are interested in the sound quality in your auralisation files, you may focus on getting
an Output Level (Max. Out level in the auralisation display within the Job list) below, but as close to 0
dB as possible, in the Convolve BRIR and Signal file table. The same applies to the Max. out in the Mix
convolved wave results into one wave file table in order to obtain the highest dynamic range. If using the
Streaming convolution option available from the main display in the Joblist, ODEON will maximize
the auralisation output level. When changing the input signal or BRIR from within this display,
you should click the Maximize Gain button to maximize the gain for the new setup.

Relative playback levels

In some cases, you'll be interested in obtaining correct relative levels (e.g. for comparisons between
different seats in a concert hall). In the case of offline convolutions, you should remember to use
the same levels (recording levels and mixer levels in the convolver and mixer windows of the
Auralisations window) in the samples to be compared; it is a good idea to use the same input signal
file to make sure that levels are the same at this point. If you wish to compare across different
rooms, you should also be aware of the source gain you are using in each room. If using the Streaming
convolution available from the main display in the JobList, ODEON will maximize the auralisation

109
output level automatically. Therefore, if you wish to compare different rooms, you should make
sure to set the Gain in the Streaming convolution window to the same value as in the other rooms.

Absolute playback levels for headphone auralisation

Setting the level to an absolute level - so the subject presented to the auralisation sample
experiences the same level as would have been the case in the real room - is tricky as it involves
every part in the signal chain.

To obtain a reasonable correct level, a first approach is to adjust the auralisation output against
levels - of some kind of sound - in the room in which you actually are. If you are simulating a voice,
try to compare the level of the playback with the level of somebody speaking in your room. This
method should make it possible to make a rough adjustment - and it is certainly better than none.

A more precise method is to use the calculated SPLA as a reference (if it is calculated at an absolute
level):
• Present the auralisation signal over a loudspeaker in the room in which you are sitting.
• Measure the sound pressure level in the room at the position where you will be listening, and
adjust the output level of the loudspeaker-amplifier until the measured LeqA corresponds to the
calculated level. At this point you have a physical reference level, which can be used for
calibration of your auralisation playback level.
• Switch between playing your auralisation sample over headphones, and over the loudspeaker,
while adjusting the level of the auralisation playback until the levels are the same.

This method is somewhat inspired by the old Barkhausen method for measuring loudness level in
Phon, and should at least in principle allow perfect calibration of the level (the resulting level being
within one subjective limen).

7.8 Input/Output Signals for Auralisations

Input signals for auralisation – anechoic recordings
For auralisation you will be using input signals to convolve with the calculated BRIR's. Usually,
the signals will be anechoic signals although it may also be other types of signals, e.g. if you are
simulating an ordinary stereo setup in a room you will probably be using a commercial stereo
recording. The input signals to be used with ODEON are stored in uncompressed wave files,
having the well-known .wav extension. Most, if not all, resolutions are supported, so files edited by
programs such as Audacity or Adobe Audition can be used without any conversion being required:
• 8, 16, 24, 32 bit PCM.
• 32 bit IEEE Float.
• 8, 16, 24, 32 bit PCM, Extensible (tells the number of significant bits).
• 32 bit IEEE Float, Extensible (includes some additional info which is not used for input by
ODEON).

To be compatible with the HRTF’s supplied with ODEON the wave files should always be at a
44100 Hz sampling frequency.

Note: 8 bit resolution is not recommended for normal use, but can be useful for education
purposes in order to demonstrate dithering. Outputs created with the Apply dither and noise shaping
disabled will sound horrific in the 8 bit resolution and output quality improves greatly when
dithering is enabled. The improvement in quality caused by dithering is less audible at higher
bit resolutions – but still important.
Remember: When the output level is higher than 0 dB undesired clipping will occurs. In this case you
have to reduce the recording level (Rcd. lev column) or the Overall Recording Level in the Auralisation Setup.

110
The ODEON program comes with a few anechoic samples, which are installed to the
\ODEON...\WaveSignals\ directory. If you wish to extend the library of input signals you should put
your new signal files here or in a subdirectory to this directory, e.g.: \ODEON...\WaveSignals\English voice\
or \ODEON...\WaveSignals\NoiseSignals\.

A few audio albums containing anechoic recordings are commercially available, namely the
Archimedes album (1992), which contains some recordings of solo instruments and the Denon
album (1988) which contains (semi) anechoic stereo recordings of orchestral music.

A growing number of anechoic recordings are available from the Odeon home page
https://odeon.dk/downloads/anechoic-recordings/. Many of the files come in the format of Wave
pack installations or OdeonZipArchives which installs anechoic recordings, directivity files and
room models which includes the complete setup of room, materials, sources, receivers, jobs,
anechoic recordings etc. ready for simulation.

Output signals
The output signals from all binaural auralisation are stored in two channel wave files and will have
the same leading name as the room. The result files, being in the wave format, makes it easy to edit
and publish the results, e.g. on the Internet or in Microsoft Power Point presentations. It is possible to
drag and drop the file directly from the tables in the Job List and Auralisations tables into e.g Microsoft
Power Point or into a folder in the Windows Explorer.

If you need to manage many of the auralisation output files then it is an advantage to know the
naming convention of these. The binaural impulse response files have the extension .Jnnnn.Wav
where nnnn refers to the relevant job number. The wave files created as results from the Convolve BRIR
and Signal file table, will have the extension .ConvAuralnnn.Wav where nnnn refers to the row in the table
(Conv. no.). The wave files, created as results from the Mix convolved wave results into one wave file table,
will have the extension .MixAuralnn.Wav where nn refers to the row number in the table (Mix. No.).

The output from surround auralisation follow rules similar to those of the binaural ones. The
impulse responses are stored in wave files following the WaveFormatExtensible format where a signal
is available for each loudspeaker/channel in the specified surround setup. The impulse response
has the extension .SurRoundnnnn.Wav where nnnn refers to the relevant job number. The convolved
files have the extensions .ConvSurRoundAurannlnn.Wav where nnnn refers to the row in the table (Conv.
no.) and vice versa for the mixed files. These files should be playable using most applications for
for audio play back e.g. the Windows Media Player.

111
 Room Acoustical Parameters
ODEON derives a large number of acoustics parameters from simulations as well as
measurements. Many of them are calculated according to the ISO standard 3382, part 1, 2 and 3 for
room acoustics, for Performance spaces, Ordinary spaces and Open-plan offices respectively (ISO 3382-
1, 2009).
This chapter will shortly describe the most common of these parameters, what they are used for,
and the theory behind. This manual will not cover the use of the individual parameters in depth.
Recommended values given for the parameters should only be considered as a first suggestion.
Besides using the default parameters, since ODEON 11 you can define your own ones or edit
existing ones in the Room Acoustic Parameters List .

8.1 Decay curve

Reverberation parameters in ODEON are related to the decay curve. This is the curve that shows
how sound energy decays in time, after a continuous noise source has been switched off. A decay
curve is derived from the impulse response, which includes all information regarding reflections
and absorption at a particular receiver position. Since each subsequent reflection becomes weaker
due to sound absorption from the surfaces in the room and from the air, a decay curve has always
a negative slope. Decay curves in ODEON are calculated for eight octave bands (63Hz to 8000Hz)
separately. The following figure shows an example. Here the overall level, as shown on the left
axis, is negative, which simply means that the sound power of the sound source used in this
example is very low. For simulations, where background noise is inherently absent, there is
absolutely no problem in having very low (e.g.: negative) levels for the decay curves. Since many
simulated parameters only depend on the slope of the curves and not the actual levels, results will
be reliable. Some room acoustics parameters do rely on ‘actual’ levels and these will be mentioned
explicitly further below.

Remember: For reverberation times and many other room acoustic parameters, the actual level of the
decay curve is not crucial. The parameters depend on the slope and the shape of the decay. Therefore,
even negative dB values are accepted, which is the case if a very weak sound source has been used in
the Source-Receiver list (e.g. the default Omni.SO8 source).

112
8.2 Reverberation Time and Early Decay Time
Reverberation time is the most frequently used parameter in room acoustics. Some acousticians
rely only on reverberation time when evaluating room acoustics, but for most cases this is not
sufficient. The reverberation time is defined as the time it takes for a sound to decay by 60 dB after
the sound source has been switched off. This is noted as T60.
The most common experience of reverberation time in a space is when making a fast hand-clap.
This corresponds to a loud sound source which has been ‘turned on’ and ‘off’ almost momentarily,
leaving a decaying tail of sound which often takes one or a few seconds to cease. Reverberation
time can reveal information about the size of a room and absorption in a room. Large rooms with
sound reflective surfaces have longer reverberation times, while small rooms with sound
absorptive surfaces have short reverberation times.

When measuring reverberation time in practice there is always a considerable level of background
noise in the recording (ambient acoustic noise, electric noise in microphone, electronics, cables, etc.)
which reduces the range of 60dB required to derive T60. Therefore, reverberation time is calculated
based on smaller decay ranges (10 dB, 15 dB, 20 dB and 30 dB) instead, which are well above the
background noise level. However, the corresponding time it takes for each decay range is always
multiplied by an appropriate factor to extrapolate the time for 60 dB decay. Calculating
reverberation time in this way results in multiple different numbers for reverberation time: The
following graph shows the process:

• Early Decay Time (EDT) is obtained from the initial 10 dB drop of the backward-integrated decay
curve, by applying the best-fit linear regression line between 0 and -10 dB. Assuming that decay
is linear, extrapolation of 10 dB range to 60 dB requires multiplication by 6. In the example
above, the time elapsed for 10 dB decay is 0.2 s. Therefore EDT=6*0.2=1.2 s.
Early Decay Time has to do with the early sound reflections only. If a sound source is
continuous, e.g. speech or music, a lot of the reverberation of a sound incident will be masked
by the next sound incidents. Thereby the late part of the reverberation will become unnoticeable.
Early decay time is related to the first -10dB and can be a better parameter to investigate the
experienced reverberation of continuous sound sources than Reverberation Time is.

113
 Since the early part of the decay is dominated by the direct sound and it is much steeper than
the rest, EDT is usually shorter than the rest type of reverberation times.
• Reverberation time T15, obtained applying the best-fit linear regression line between -5 and -20
dB (not shown in the graph above), and then multiplied by 4.
• Reverberation time T20 is similar to T15. The linear regression line is found from -5 to -25dB and
then multiplied by 3. In the example above T20=3*0.6=1.80s.
• Reverberation time T30 is similar to T15. The linear regression line is found from -5 to -35dB and
then multiplied by 2. In the example above T30=2*0.93=1.86s.

For simple rooms, with evenly distributed absorption, the different values of reverberation times
can be very close to each other, since the decay is almost a straight line (in decibels, therefore a
logarithmic scale). For coupled spaces and other complex rooms, the decay curve might change the
slope through time, therefore reverberation time can be quite different depending in which part of
the energy curve has been used for evaluation.

Guidelines for simulating Reverberation time

According to the definitions above, the various reverberation times T15, T20 and T30 do not take into
account the initial part of the response, i.e., between 0 and -5 dB. This means they are not affected
by the direct sound from the source, but mostly from the late reflections, which become
progressively more diffuse over time. Consequently, we expect that the spatial variation of T15, T20
and T30 is much smaller comparing to other room acoustic parameters that depend on the direct
sound.
For example, although EDT is defined as a variation of reverberation time, its behaviour is much
different than the rest reverberation times. Since it is evaluated between 0 and -10 dB on the decay
curve, it is heavily affected by the direct sound and the early reflections, so that its fluctuations
throughout the space are significantly higher.

Spatial distribution
Looking at the following example of grid calculations in ODEON, variation of reverberation time
T30 in a regular volume is not as big as of Early Decay Time (EDT).

Left: Grid calculation of T30 on the floor and audience surfaces at 1000 Hz. The values vary mostly between 1.8 and 2.4 sec
(app. 0.6 sec range). Right: Grid calculation of EDT on the floor and audience surfaces at 1000 Hz. The variation is greater,
between 1 and 2.25 sec (app. 1.25 sec range).

Reverberation time is considered a global parameter, which means that it is highly affected by the
room as a whole and not by local reflections. Therefore, it is preferred to be reported as an average
over a number of source-receiver combinations, instead of specific receiver locations. The standard
provides suggestions for the number of sources and receivers to be included in the averages,
according to the desired level of precision in the simulations (or measurements).
The ODEON’s Multi Point Response in the Joblist (see Section 5.2), is probably the best and most handy
tool to simulate or measure reverberation time, according to (ISO 3382-1, 2009) and (ISO 3382-2,

114
2008). You can use other simulation tools, like the Global Estimate (see Section 5.1) or the Grid Response,
but none of them can be entirely consistent with the standard, as will be seen further below.

Remember: It is generally not recommended to report reverberation times T15, T20 and T30 at specific
receiver positions, but as an average obtained from several source-receiver combinations.

Source-receiver positions
According to (ISO 3382-2, 2008) the sources can be located at a normal position of human speakers,
loudspeakers etc. according to the expected use of the room. In small rooms such as domestic
rooms or when no normal positions exist, one source position should be in a corner of the room.
- Receiver (microphone) positions should preferably be at least half a wavelength apart. For
frequencies above 63 Hz, this corresponds to 2.7m. However, considering that the frequency
range for speech starts from 125 Hz, this distance becomes 1.4m. That is why the standard gives
the suggestion of 2m for the usual frequency range.
- The distance from any receiver position to the nearest reflecting surface, including the floor,
should be at least a quarter of a wavelength, i.e. around 1 m for the usual frequency range.
- Symmetric positions should be avoided.
- The receiver should be at the following minimum distance (in meters) from the source:

𝑉
𝑑𝑚𝑖𝑛 = 2√
̂
𝑐𝑅𝑇
The formula is particularly useful for measurements and not simulations. If we use the
simplified Sabine equation that relates reverberation time with volume 𝑉 and absorption area
𝐴, 𝑅𝑇 = 0.16𝐴/𝑉, we can rewrite the formula for minimum distance as:

𝐴
𝑑𝑚𝑖𝑛 = 2√
0.16𝑐

where 𝑉 is the volume of the room in m3 and 𝑅𝑇 ̂ is an estimation of the expected reverberation
time in sec. Both values can be easily obtained from the Quick Estimate in the Material list.

Obtain RT from Quick Estimate. The reverberation times given from Sabine and Eyring formulas provide a sufficient
estimation for calculating the minimum distance between source and receivers. These values are calculated based on the
source location (at the bottom of the window), but the variation is minimal. It is only used by ODEON to detect the volume,
where the source is located, to ensure that surfaces from other potential coupled volumes are excluded from the calculation.
See more in Section 5.1.

115
Obtain V from Quick Estimate. The second tabsheet of the Quick Estimate window provides a sufficient estimation of the
room volume, which can be used when calculating the minimum distance from the source.

According to the guidelines above, neither the Global Estimate or the Grid Response in ODEON satisfy
all necessary distances between sources, receivers and reflecting surfaces. The Global Estimate
provides calculation for the entire volume of the space (not only close to the human-receiver
height), while the Grid Response - although it can be set at a realistic height - can include receivers too
close to the source or walls. Therefore, both types of calculation should be use with care. Instead,
to have full control over the location of sources and receivers the Multi Point Response calculation
should be used.

The Table below shows the minimum number of required source-receiver positions, according to
the precision desired for the final average of the reverberation time. These numbers are taken
directly from (ISO 3382-2, 2008) which has been developed for measurements. We can follow the
same guidelines in simulations since the only difference is the lack of ambient (background) noise.
the lack of ambient noise can only improve the simulation results.

Survey Engineering Precision

Source-receiver ≥2 ≥6 ≥12
combinations
Number of sources ≥1 ≥2 ≥2
Number of receivers ≥2 ≥2 ≥3
When the result is used for a correction term to other engineering-level measurements, only 1 source-position and 3
microphone-positions are required.
A number of source-receiver combinations has to be included for each of the three levels of accuracy. There is freedom to
choose how many sources or receivers to include in the combinations, but the minimum number of each must be respected.
For example, for a minimum of 12 combinations in the Precision method only the following cases can be acceptable: 2
Sources – 6 Receivers, 3 Sources – 4 Receivers, 4 Sources – 3 Receivers.

We apply the same principles in ODEON simulations, and therefore the Room Setup has been
equipped with the buttons Survey, Engineering and Precision, in order to make the connection to the
measuring standard. However, these settings only affect the number of rays and the rest calculation
parameters and not the number of sources or receivers. They are used as an indication of simulation
accuracy. See more about the different parameters in the room setup in Chapter 10.

116
In the following picture an example of Engineering calculation is illustrated with 2 Sources and 3
Receivers, with some of the critical distances indicated.

𝑑𝑚𝑖𝑛

Example of engineering calculation with 2 sources and 3 receivers. All distances from the nearest walls are >1m and all
distances between receivers are >2m. The minimum distance 𝒅𝒎𝒊𝒏 between source and receiver ensures that the receiver is
not in the near field of the source, but in the reverberant field. Thus, it is not heavily affected by the direct sound.

Occupancy state
In many rooms, the number of people present can have a strong influence on the reverberation
time. On a general basis, reverberation time measurements and simulations should be made in a
room without people. However, the absorption from people can always be included if it is
necessary for a particular simulation, as long as it is clearly written in the report.
Specifically for performance spaces, the ISO standard (ISO 3382-1, 2009) defines three states of
occupancy that can be used as a reference.

Unoccupied state
State of a room prepared for use and ready for speakers or for performers and audience, but
without these persons being present, and in the case of concert halls and opera houses, preferably

117
with the performers' seats, music stands, heavy instruments (percussions, piano etc.) and generally
any immobile equipment.

Studio state
State of a room occupied by performers or speakers only and without an audience (for example,
during rehearsals or sound recordings) and with the number of performers and other persons such
as technicians corresponding to the usual number.

Occupied state
State of an auditorium or theatre when 80 % to 100 % of the seats are occupied by audience. In cases
of extraordinary occupancy (for example an orchestra larger than usual, or a choir/standees are
present), the extra absorption introduced shall be included in the simulations.

Coupled volumes
Often, spaces are connected via corridors or openings, and although they are not physically fully
separated by walls or doors, it is not recommended to study them as one volume. According to
(ISO 3382-1, 2009) such coupled spaces are likely to show areas with differing reverberation times.
Therefore, they must be investigated and measured separately. In other words, both sources and
receivers must be inside the same space. There is not much benefit in having the source in one of
the coupled spaces and the receiver in the other.
In the following open-plan office example we can identify two large spaces, connected via a smaller
one. The source is located at the lower space. The most correct calculation/measurement of
reverberation time would be at the receiver positions marked by the dashed-green rectangle, while
the least correct would be at the receiver positions marked by the dashed-red rectangle at the upper
space. Calculating reverberation time in the middle space would also be acceptable and accurate
enough, as the common opening is relatively large comparing to the volume of the space.
Moreover, there is direct contribution from the source inside the middle space, which makes the
decay response quite accurate.

118
When simulating/measuring reverberation time in coupled spaces, make sure the source is at the same space
as the receivers, and try to split the calculation for the different spaces. Adjacent spaces, with large connecting
openings with the main space can be included at the same calculation. In the example above, with the source
located at the lower space it makes sense to calculate reverberation time around the areas marked with the
dashed-green rectangle.

Impulse Response Resolution

The energy of all the reflections received at the receiver point is collected in histograms in regular
time intervals, which are specified in the Room setup>Impulse response resolution. After completion of the
response calculation, early decay time and rest reverberation times are calculated according to the
procedure described above.

Problems Calculating Reverberation Time

Many times reverberation time fails to be calculated, leaving *.** values in the Energy Parameters
tabsheet or empty bars in the Parameter bars Tab Sheet.

119
 Warning: The most common reason for failing to obtain results is that the impulse response
length in the calculation is too short. Try going to Room setup |Impulse response length and make it
longer. Typically, 2/3 of the expected reverberation time is enough.

Another problem can be number of late rays. If the room is complex, consists of coupled spaces or
is outdoor, there will be a need for a higher number of rays. Go to Room setup |Number of late rays and
increase it. You can always look at the Decay curves to evaluate if there are enough rays. A smooth
slope means enough rays, where one with drastic drops could be a sign (though, not necessarily)
of too few rays.

Curvature, C
Curvature is an important parameter indicating how straight the decay curve is. It is defined as:

𝑇30
𝐶 = 100 ( − 1) (%)
𝑇20

Ideally, decay curves should be as straight as possible, corresponding to an exponentially decaying,

diffuse field. In such a case, T20 and T30 would be equal and the curvature would be 0. In real life,
the decay curve is rarely straight, with the early part being typically steeper than the late part. T20
is affected mostly by the early part, while T30 by the late. Values of curvature below 10% are
perfectly acceptable for most practical applications, while values higher than 15% are not
recommended. In such cases, the late part of the decay has significantly shallower slope than the
early part, typically because of flutter echoes between parallel hard walls and uneven absorption
in the room. The most common example is absorption concentrated on the ceiling, while the walls
remain hard. This results in a 2-dimensional sound field (between the parallel hard walls) which is
illustrated in the double-slope decay curve. When a high curvature value is present, the number of
late rays in the Room Setup needs to be increased dramatically, well above the Precision setting.
Otherwise, the results become highly unreliable and values of room acoustic parameters fluctuate
with increasing number of rays, until convergence is achieved.

In the following example the bended curve problem is illustrated. The dashed lines roughly
indicate the slopes that correspond to T20 and T30 for the low-mid bands. The calculated Curvature
is shown below and it is well above 15%. In this example 22000 late rays were used, instead of the
recommended 16000 by the Precision setting in the Room Setup.

120
 5 dB lower from peak

Regression line interpolated between

the points of the decay curve.

T30

T20

The following table aims to give some general guidelines when evaluating curvature in
simulations.

Curvature Situation Number of late Rays Comments

C(%) needed in the Room
Setup
<10 Acceptable bending of Engineering – Precision Results are reliable and
decay curve. setting. converge with an increasing
number of rays.
10-15 Bending of decay curve At least 2 times the You need to increase the
starts to become an Precision setting. number of rays beyond the
issue. Precision setting to achieve stable
results.
>15 Extreme bending of At least 10 times the Simulations become unreliable
decay curve that should Precision setting. and convergence might only be
be avoided. seen after using an enormous
number of Rays.

121
8.3 SPL, G, C80, D50, LF80, STearly, STlate and STtotal
Multiple acoustical parameters have been defined through time, to investigate different influences
of the room acoustics on the subjective experience of sound. ODEON includes a range of these, and
you can add your own based on the same principles. Many of these parameters have to do with
sound energy arriving at the receiver positions at different time intervals. For example, the
parameter Clarity (C) is based on the fact that for some music and for speech, the sound energy
arriving before 50 or 80ms has the most influence on how clearly the sound is perceived, compared
to the energy arriving later.

When derived in ODEON the energy of each reflection is added to the appropriate term in the
formula of all the energy parameters, according to its time and direction of arrival. After the
response calculation, C80 (Clarity), D50 (Definition), Centre Time, SPL (Sound Pressure Level), LF 80
(Lateral Energy Fraction), STearly, STlate and STtotal are derived.

Explanation of formulae
Only a few common parameters that are available in ODEON are listed here. You can find the
definitions of parameters from within the Room Acoustic Parameter List in ODEON where the formulas
are listed, and you can also enter new parameters.

In the following formulae, Ea − b is the sum of energy contributions between time a and time b after
the direct sound, time t is the end of the calculated response, and t is the angle between the
incident direction and the axis passing through the ears of a listener, for the reflection arriving at
time t.
It is worth noticing that parameters relying on the Ea − b term assume that time a and b are well
defined - however this is not the case when there is no direct line of sight between source and
receiver. It makes results questionable in simulations as well as in real measurements when direct
sight between source and receiver is not present.

SPL (Sound Pressure Level)

The SPL is the level of the sound pressure. It is a parameter often used in regulations for e.g.
industrial noise levels, since high levels damage hearing. It is also used for regulations on sound
levels in living quarters or concerts, for the consideration of neighbours.

When using SPL in ODEON, it is important that the sound source has the correct Sound Power,
since an arbitrary Sound Power at the source will give an arbitrary SPL at the receiver positions.
Measurements in real rooms using ODEON will normally give an arbitrary SPL level, because the
hardware used is not capable of obtaining exact SPL levels. It can, however, give relative SPL levels
compared to each other. When the source power is 31 dB per octave then SPL equals the G
parameter (Strength) assuming an omni directional source.

SPL = 10 log( E0− ) (dB)

G (Strength)
The G parameter (Strength) is defined in (ISO 3382-1, 2009). It helps investigate how much the
acoustics/reflections of the room contributes the level of the source. In a very reflective space/room
there will be a high contribution, in an absorptive space there will be a low contribution. The
reference level of 0 dB is obtained in free field at a distance of 10 metres from an omni directional
point source. This is obtained by normalizing the Sound Power of an OMNI directional point
source to 31dB per octave. Odeon will only calculate the G parameter if the job in Job List:

122
 • Contains only one active source
• The source is a point source
• The directivity file name contains the substring omni (case insensitive)

In this case Odeon will automatically normalize the source power to 31 dB per octave for the
calculation of the G parameter(s) – see below.

If you want to measure G using ODEON in a real room, you need to calibrate the measuring
system, software, and hardware, using one of the different calibration methods available in
ODEON and you should use an omni directional source.

Gearly and Glate

Gearly and Glate are two additional strength parameters used in auditorium acoustics to see how
much the room contributes to the early and late part G. The early G includes the energy arriving
within the first 80 ms and the late G the energy arriving after the first 80 ms, relative to direct sound.
Details given for the G parameter also apply to Gearly and Glate.

C50 , C80 and C7 (Clarity).

Early reflections arriving before 80 ms after the direct sound contributes to the clarity of symphonic
music whereas sound arriving after 80 ms has a “blurring” effect. The (ISO 3382-1) states that C50
is appropriate for speech and C80 for music. It should be noted that much acoustical research has
been made for classical music, but for rhythmical music (pop, rock, etc.) it may not apply in the
same way. C50 investigates the relation between sound energy before and after 50ms. For C 80 it is
80ms. ODEON also includes C7, a parameter that investigates the relation between what may be
expected to be direct sound (before 7ms) compared to reflections. The C 7 is mainly for
measurements, since direct sound in simulations will be theoretically ideal and will arrive at 0s.

E 
C80 = 10 log  0−80  (dB)
 E80− 

D50 (Definition)
Definition is very close to being the same as clarity, but both exist due to historical reasons.

E 0−50
D50 =
E 0−

Ts (Centre time)
Relates to perceived definition, clarity, or the balance between clarity and reverberance, as well as
to speech intelligibility.


 tE t
Ts = 0
(ms )
E 0−

LF80 (Lateral Energy Fraction)

LF is a parameter for the spaciousness of the room. A room is acoustically very spacious if it
makes a sound source be perceived as “wider”.
80

E t cos 2 (  t )
LF80 = t =5

E0−80

123
The LF80 parameter has a high correlation with the apparent source width (ASW) as shown in
(Bradley, et al., 1995).

Lj (Average)
This parameter is suggested in (Bradley, et al., 1995) and is included in (ISO 3382-1, 2009).
Lj(Average) has a very high correlation with the subjective parameter Listener envelopment (LEV).

 1 N1000 Hz  

Lj ( Average) 80 = 10 log    Et cos 2 (  t ) (dB)
 4 N125 Hz t =80 

An alternative name for this parameter could have been “Late Lateral G”, because the level is
normalized in the same manner as for G, Gearly and Glate . Details given for the G parameter also
applies to Lj.

Example of target values

Depending of the purpose of a room, the target values for the acoustical parameters will be
different. Auditorium acoustics as in Concert Halls, Opera Halls, Multipurpose halls, etc. are dealt
with in (Beranek, 1996) and (Barron, 1993), where halls from all over the world are presented along
with a judgement of their acoustics, supplied by a guideline for design. Short guidelines on values
for Clarity and G in concert halls, based on some simple design parameters such as width, height,
floor-slope, etc. are given in (Gade, 1997). Below some recommended values for room acoustical
parameters are shown:

Objective parameter Symbol Recommended Subj. limen

(symphonic music)
Reverberation time T30 1.7 - 2.3 seconds 5%
Clarity C80 - 1 to 3 dB 1 dB
Level rel. 10 m free field G > 3 dB 1 dB
Early Lateral Energy Fraction LF80 > 0.25 5%
Early Support STearly > -13 dB
Total Support STtotal > -12 dB
Recommended values for objective room acoustical parameters (ISO 3382-1, 2009) in large music rooms with
audience, according to Gade (Gade, 2003). Subjective limen as given by Bork (Bork, 2000) and Bradley (Bradley,
1986).

8.4 Stage Parameters

Stage parameters refer to how the acoustics influence the experience for the musicians playing on
a stage. The relation between the direct sound and reflection from the surfaces affects how well can
they hear themselves and each other.

The parameters are called Early, Late and Total Support, and are described more thoroughly in
(Gade, 2003):

Stage parameters are calculated as a part of the Single Point response (Auditorium and Combined
versions only). If the job only contains one active source, the active source is a point source and the
distance between receiver and source is approximately 1 metre (0.9 to 1.1 metre).

Early Support or ST1:

STearly or ST1 is used as a descriptor of ensemble conditions, i.e. the ease of hearing other members
in an orchestra.

124
 E 20−100
ST early= (dB)
E 0−10

Late Support:
STlate describes the impression of reverberance.

E100−1000
ST late= (dB)
E 0 −10

Total Support:
STtotal describes the support from the room to the musician’s own instrument.

E 20−1000
ST total= (dB)
E 0 −10

8.5 IACC – Inter-Aural Cross Correlation

Since the hearing process is binaural, the spatial impression in a concert hall can be quantified by
measuring the inter-aural cross correlation between the left and right ear canal (ISO 3382-1, 2009).

IACC in math
The Inter-Aural Cross Correlation is expressed as a coefficient from 0 to 1. First, the normalized
inter-aural cross correlation function is calculated:
t1

 p (t )  p (t +  )dt
l r

IACFt1 ,t2 ( ) =
t2
t1 t2

p (t )dt  pr (t )dt
2
l
t2 t1

where:
pl (t ) is the impulse response at the entrance to the left ear canal.
pr (t ) is the impulse response at the entrance to the right ear canal.

Then, the inter-aural cross correlation coefficient is calculated as:

IACC t1 ,t2 = max IACFt1 ,t2 for −1 ms    +1 ms .

IACC in ODEON
To simulate IACC in ODEON you need a virtual dummy head. Therefore, these parameters are
only calculated as part of the Single point response. In order to calculate IACC correctly a neutral
dummy head without localization enhancement is needed. To create a new HRTF filter without
localization enhancement:
• Open Tools > Create filtered HRTF.
• Choose Subject_021Res5deg.ascii_hrtf.
• Press ok (make sure "Apply localization enhancement" is NOT on).
• Go to the "Auralisation setup" in the top-menu bar.
• Checkmark "Create binaural impulse response file".

125
 • For the HRTF choose:
Subject_021Res10deg_SRate44100_Apass0,50_Astop40,00_BOvrLap100%_PPrHRTF256 .
• Delete everything in Headphone (the area will turn red).
• Close window.

If this is something you do often/in many rooms, you can change the default:
• Go to Options > Program Setup.
• Choose Auralisations tab.
• For the HRTF choose:
ChoosSubject_021Res10deg_SRate44100_Apass0,50_Astop40,00_BOvrLap100%_PPrHRTF256
• Delete everything in Headphone (the area will turn red).
• Close window

8.6 STI – Speech Transmission Index

To investigate how well speech is understood for a receiver in a room, the Speech Transmission
Index, STI, can be used. It is calculated in ODEON according to (IEC 60268-16, 2011).

The STI parameter takes into account the background noise, which may be adjusted from the Room
setup. For the STI parameter to be valid, it is very important to adjust the background noise
accordingly see below.

Do notice that it is not stated in the (IEC 60268-16, 2011) standard what kind of directivity the source
in the STI measuring system should have. So, if using a source with a different directivity in
simulations than the one used in the real measurements, results may not be comparable. The
subjective scale of STI is given below:

Subjective scale STI value

Bad 0.00 - 0.30
Poor 0.30 - 0.45
Fair 0.45 - 0.60
Good 0.60 - 0.75
Excellent 0.75 - 1.00

In the point response and multi-point response results, five different speech transmission indexes
are shown:

STI
This is the best-known form of STI is derived for an average gender-independent voice spectrum.
This basic STI parameter can be calculated using relative source power levels e.g. you may set the
power of the source to 0 dB and still get meaningful results If using relative source power levels it
is important though that the background noise (SPL) is set in a relative level if relative source gains
are used. To exclude the background noise entirely from calculations, leave the background noise
levels at -99 dB in the Room setup.

STIMale and STIFemale

These are STI values are adapted for male and female speakers, according to their speech spectrum.
Contributions of each Octave Band to the STI as well as the information overlap from one band to
another (masking), is taken into account. These two STI values also include compression effects in
the human hearing which decreases their values at high levels (in particular relevant for Voice
alarm system). Because calculations depend on the actual sound pressure levels that occurs at the
receiver, the source power levels of the sources must be the actual ones. In similar manner the

126
levels of the background noise in the Room setup must be the actual levels as they are measured in
the real room.

STIExpected
The expected STI value is calculated on a theoretical basis, taking into account the reverberation
time of the room and the background noise, but not the explicit shape of the impulse response. In
other words, the STIExpected is the value of STI that would be expected in a diffuse sound field with
the given reverberation time and noise level. Comparing actual STI values with the expected ones
can indicate if STI has been improved – the actual values should be higher than the one expected
in a diffuse field.

RASTI
An abbreviation of Rapid STI that is considered an obsolete parameter in the revised (IEC 60268-
16, 2011) standard. It is based only on two Octave Bands: 500 Hz and 2000 Hz and provides a faster
and simplified calculation of the STI. Nowadays, in simulation and in measurements using the
sweep method as in ODEON, speed is no longer an issue.

8.7 DL2 – Rate of Spatial Decay

ISO 14257 parameters (ISO 14257, 2001) can be calculated as part of the Multi point response from within
the Job List. In order to calculate ISO 14257 parameters in ODEON 16, a point source needs to be
defined using the Point source editor in the Source-Receiver List. This point source must be assigned a
Directivity pattern which contains the text OMNI in its name. This source and no other source must
be activated in a job and a number of receivers with different distances from this source must be
defined. If the above conditions are not met, these ISO parameters are not available in the Multi point
response display.

Remember: The ISO 14257 parameters need certain conditions to be met in order to show up in the
Multi point response display: Only one OMNI source active and a number of receivers placed at different
distances.

The rate of spatial decay DL2 is the decay of sound pressure level per doubling of distance. The DL2
parameter is intended to characterise the acoustic performance of workrooms.

DL2 is calculated in ODEON according to (ISO 14257, 2001). The values to be expected for the DL2
parameter are, according to (Ondet, et al., 1995): 1 to 3 dB for reverberant rooms and 2 to 5 dB, for
ideally treated rooms. The design criterion for DL2 is set to 3.5 dB or better according to (ISO 11690-
1, 1996).

The DL2 parameter is calculated as part of the Multi Point response. If the job only contains one active
source, the active source is a point source, and if more than one receiver is defined, the distance
between the source and the receivers is not the same for all receivers. Please notice that one
misplaced receiver may ruin the entire DL2 calculation, thus it is a good idea to check the receiver
positions - or even better - check the individual results of the Multi Point calculation.

DL2 is stated for the frequency bands 63 Hz to 8 kHz, and DL2,Co. is the A-weighted Rate of Spatial
Decay for the frequency bands 125 Hz to 4 kHz. For DL2, as well as DL2,Co the correlation coefficients
are calculated. If the correlation coefficients are low, this may indicate bad locations of source
and/or receivers, however it may also indicate a very low damping in the room (the Spatial Decay
Curve being almost horizontal).

127
The measuring points (Receivers) and the source position are of course essential to the DL 2
parameters and should follow (ISO 14257, 2001). As an example, a path of receivers may be chosen
in the following distances from the source (using logarithmic increment):

1, 2, 4, 5, 6.3, 8, 10 metres

The positions should also follow the standard with respect to distance from floor and reflecting
surfaces. ODEON will use all the receivers defined in the Receiver list. The Multi point response display
will calculate and display values for DL2 based on all the receivers visible in the 3D Source and Receivers
tab in the Multi point response. Make sure to create a group with relevant receivers in this display – or
simply deactivate irrelevant ones.

8.8 ISO 3382-3 (Open-plan offices) parameters

ISO 3382-3 parameters (ISO 3382-3, 2012) can be calculated as part of the Multi point response from
within the Job List. In order to calculate ISO 3382-3 parameters in ODEON 16, a point source needs
to be defined in the Point source editor. This point source must be assigned the Directivity pattern:
ISO3382-3_OMNI.SO8. This source and no other source must be activated in a job and a number of
receivers must defined within a distance of 2 to 16 metres from this source. If the above conditions
are not meet these ISO parameters are not available in the Multi point response display.

Remember: The ISO 3382-3 parameters need certain conditions to be met in order to show up in the
Multi point response display: only one ISO382-3 OMNI source active and a number of receivers placed at
distances between 2 and 16 metres from the source.

The ISO standard 3382-3 describes a range of parameters to investigate the influence of room
acoustics in open plan offices. There are two application notes on our website explaining how to
work with these parameters (currently located at https://odeon.dk/learn/application-notes/). The
following parameters are supported:

Spatial distribution of STI

This curve shows the decrease of speech transmission index as a function of the distance from the
source.

Distraction distance - rD
It is the distance from the sound source where the speech transmission index falls below 0,50. Here
we focus on the receiver and his/her ability to concentrate when somebody talks at a distance.
Further than the distraction distance, concentration and privacy start to improve dramatically for
the receiver.

Privacy distance – rP
It is the distance from the speaker where the speech transmission index falls below 0,20. Here we
focus on the talker and how well his/her speech is protected from being spied by the receivers in
the office. Further than the privacy distance, concentration and privacy are experienced very much
the same as between separate office rooms.

Remember: We want both Distraction distance and Privacy distance to be as short as possible, but for
different reasons. In the first, we want to ensure that the receiver is not distracted by people talking in
the office. In the second we want to ensure that the talker (source) cannot be spied by the receivers.

128
Spatial decay rate of A-weighted SPL of speech – D2,S
This curve shows the decrease of the A-weighted SPL per doubling of distance from the sound
source. The sound source emits noise with the sound power spectrum of normal speech (directivity
pattern: ISO3382-3_OMNI.SO8).

A-weighted speech at 4 meters

Nominal A-weighted SPL of normal speech at a distance of 4m from the sound source. These levels
are summed up over the octave bands from 125 to 8000 Hz.

A-weighted background noise

The quantity shows the level of background noise summed up over the octave bands from 125 to
8000 Hz.

8.9 Edit/Make Room Acoustic Parameters

In ODEON, the user is able to edit the room acoustic parameters calculated in the point response -
or define their own. Get more information on how to do this by using the context-sensitive help
(F1 shortcut) while having the Room Acoustic Parameters list open.

In the Room acoustic parameter list, you can select which room acoustic parameters will be visible in
results, what range graphs will display for different parameters, change the parameters or make
new ones.

In the unexpanded view, you can see which parameters that will be visible in the results. By
clicking the plus in the upper-right corner, you go to the expanded view. Here you can change
visibility the parameters, edit parameter definitions and create new ones.

The more basic user options are:

• Change the name of a parameter
• Make a parameter active/inactive by using the ‘visible’ option.
• Set parameters such as T30 and Centre Time, which have a natural origin, to be displayed
in graphs with ordinates starting at zero.
• Set a manual range of values in the grid-display. If not set, the range will be automatic for
most parameters, and therefore changing for each calculation.
• Choose if the Jnd is in %. The default settings will be in reference to literature, so only
change it if needed.
• Choose which of the parameters are measured at all.

The advanced user options are:

• Define the Type of a parameter: Reverberation, T30, SPL, Centre Time, Formula, Dietsch,, MTI,
MTI(corrected), G, G(early), G(late) and Lj.
• Edit the formula for a parameter. How to do this is mainly described in the help file, press
F1 when in the Room acoustic parameters list.
• Make a new parameter. How to do this is mainly described in the help file, press F1 when
in the Room acoustic parameters list.

You can also:

• Save a parameter list for later use.
• Load a saved parameters list.
• Restore default parameters list.

129
 Hint! Odeon Auditorium and Combined includes a large number of predefined room acoustics
parameters and allows this list to be expanded. Depending on the type of project you may want
to entirely exclude and to make some parameters invisible by default. Use the Load parameters from
program folder to use one of our predefined pre-sets for different kinds of projects e.g. Class room,
Concert hall, Industri, Open plan office and Rock_venue. This will make the calculations results shorter and
clearer and make it faster to scroll between parameter results.

The value of energy intervals in the left side of the window can be used as a variable in a parameter
expression, when changing or making a new parameter. If you have to use an energy interval from
a specific time point t’ to the end of the response, you have to make use of the predefined whole
response length and subtract the interval from t=0 sec to t’ (e.g. E_Omni50ms_Total = E_Omni_Total -
E_Omni50ms). For a formula type, the equation can be written explicitly in the white box. A list of all
mathematical expressions available in the edit formula mode is found in Appendix H:
Mathematical Expressions.

130
 Calculation Principles
ODEON is an energy-based room acoustic modeller, meaning that sound waves are represented
by rays. A ray can be understood as a straight line connecting a source and a receiver. Reflections
are represented by image and secondary sources, where the sound is still considered as rays.
These simplifications make it possible to calculate the acoustic response in large spaces, while
maintaining a short computation time and a high level of accuracy. Phase information is not
included in the propagation of sound.
All calculations in ODEON are performed in the energy domain - not in the pressure domain.
Therefore, the acoustic differential equations cannot be solved directly as it is done in Finite Element
(FEM) or Boundary Element (BEM) methods.

9.1 What can be modelled in ODEON

The calculation methods from room acoustics that are used in ODEON are mostly for geometrical
tasks, and they are valid for spaces and frequencies where high modal overlap occurs. That is,
when there are so many modes (standing waves) that the sound field becomes smooth. Single
modes are prominent at low frequencies, in small, reverberant rooms. But fortunately, a moderate-
sized space (like a classroom) is already considered a big volume where there is high modal
overlap, so the room acoustic algorithms used in ODEON can be applied safely. The following
theoretical formula, proposed by Schroeder, gives the minimum frequency at which there is
guaranteed to be enough modal overlap that such energy-based methods can be applied:

RT
f sch = 2000
V

RT is the reverberation time in the room (inversely proportional to absorption) and V is the volume.
As an example, in a concert hall with reverberation time 2 sec and volume 600 m3, the Schroeder’s
frequency would be 115 Hz. Therefore, an energy-based calculation in the octave band 125 Hz can
be considered reliable in terms of modal overlap, since most of the frequencies in the band are
already above the Schroeder frequency. Additionally, in practice, we can accept frequencies well
below the Schroeder’s limit, especially in large halls. Experience has shown that even frequencies
lower than 50% from this limit can be reliable.
Taking into account the limitations from the energy-based (high frequency) methods, we assume
that ODEON calculations are more reliable in moderately-sized and larger rooms. As absorption
becomes higher, this size can be reduced. ODEON has been successfully used for rooms such as:

• Meeting Rooms. • Restaurants.

• Classrooms. • Shopping malls.
• Offices – open plan offices. • Multi-floor buildings with atriums.
• Auditoriums. • Cinemas.
• Concert halls. • Theatres.
• Opera halls. • Open theatres.
• Big rehearsal halls. • Train stations.
• Small scale city plans (outdoor • Airport terminals.
modelling). • Noise halls (factories).
• Cathedrals, worship spaces. • Industrial spaces.

131
9.2 Global decay
ODEON contains two methods for calculating the global decay of rooms. These should not be used
in place of calculations from sources to receivers, but are good for first or global impressions:
• Quick Estimate, which is available from the Materials List, is the fastest method allowing quick
evaluation of the effect of modifying materials. This method should be considered only as a tool
for preliminary results.
• Global Estimate is a more precise method, allowing high quality results, though still not
representing the sound experience of a receiver at a specific position, which may be slightly
different.

Quick Estimate
The Quick estimate tool estimates volume and total absorption, which is inserted in the Sabine,
Eyring and Arau-Puchades formulas to give an estimate of the reverberation time. Instead of
simply taking the areas of the surfaces and multiplying by the corresponding absorption
coefficients to obtain the total absorption in the room, ODEON also sends out 'particles' from the
source - assuming diffuse conditions and reflecting them in random directions, keeping track on
how many times they hit each surface. Surfaces that are hit very often contain higher influence of
the overall absorption coefficient of the room. Surfaces, which are not detected at all in the ray-
tracing process, are left out of all calculations - and surfaces which are hit on both sides are included
twice in the calculation. As a result, the estimated reverberation time corresponds to the sub-
volume in which the selected source is located. Do note that if a part of the area of a surface, which
is present in the sub-volume, is located outside that sub-volume (e.g. if two sub-volumes share the
same floor surface) then the area and surface estimates for the statistical calculations may not be
entirely correct. To partly compensate for that it is possible to set a volume manually.

The classical mean absorption coefficient is given by:

S  i i
= i

S i
i

Si and i are the area and absorption coefficient of the ith room surface, respectively.

The modified mean absorption coefficient as experienced by the particles is:

H  i i
' = i

H i
i

Hi is the number of hits on the ith room surface.

In ODEON, both mean absorption coefficients are inserted in the Sabine and Eyring formulae to
calculate reverberation times; the classical values are labelled Sabine and Eyring, and the values
using the modified mean absorption coefficient are labelled Modified Sabine and Modified Eyring. The
mean absorption coefficients used for the Arau-Puchades formula are derived in similar ways
except that separate values for surface hits, area, and the corresponding mean absorption
coefficients, are calculated as projections onto each of the axes of the room. The Sabine, Eyring and
Arau-Puchades formulae require a value for the room volume. This is estimated by ODEON from
the mean free path, experienced by ray tracing and using the well-known relation for the mean free
path in a 3-dimensional room:

132
 4V
l=
S

where V is the room volume and S is the total active surface area. The ray-tracing process is
carried out in order to estimate the room volume, considering scattering coefficients of 1 for all
surfaces, rather than using the coefficients assigned to the surface in the materials list, where the
mean free path formula is based on diffuse field assumptions.

The value of S is used in the sum of the areas of non-transparent surfaces, taking into account
whether one, two or indeed none of the sides of a surface are visible inside the room.

Convergence criterion
A certain number of particles must be sent out and followed around the room for a stable estimate
to be obtained. More and more particles are sent out in random directions until the value of the
reverberation time has remained within 1% for at least 50 particles. At the end of a run, the data on
how many times each surface was hit is stored. Then, if new materials are assigned to the surfaces,
the reverberation time can be recalculated instantaneously, without repeating the particle tracing.

Global Estimate
The global estimate tool calculates the global reverberation times T20 and T30 using the method
proposed by Schroeder (Schroeder, 1970), as well as the mean free path, and generates estimates of
decay curves. Particles are sent out in random directions from the source (see Chapter 9.8) and
reflected using the ‘Vector based scattering method’ (see Chapter 7.5). ODEON records the loss of
energy in each particle, as a function of time, due to absorption in room surfaces and in the air.
Summing over many particles, a global energy decay function for the room is obtained. The decay
curve is corrected for energy, which is lost due to the truncation of the decay curve. This is
analogous to an ordinary decay curve, except there is no specific receiver. The summation process
may be carried on for as many rays as desired.

Evaluating results
When the decay curve seems smooth, you can derive the results. It is often the case that T30 values
are larger than T20 values. If T30 values are lower than T20, it is likely that the used numbers of rays
were too small - in this case you can press the Recalculate button. If the reverberation times are shown
as *.**, the Impulse response length defined at the Room setup page is too short to derive the reverberation
parameter.

9.3 Response from Sources to Receivers

This section describes the room acoustic simulation methods used to predict the (impulse) response
from sources to receivers within from the JobList. The results can be shown as Single Point Responses,
Multi Point Responses and Grid Responses. In ODEON Auditorium and Combined, the predicted response can
be used for auralisation of any sound clip (see Chapter 7). These simulation methods aim to mimic
real impulse response measurements, although they all operate in energy domain, while real
measurements operate in pressure domain (see sec. 9.1). As a result, impulse responses in ODEON
are inherently positive as a function of time, corresponding to the squared impulse response of a real
sound pressure measurement. Real measurements show both positive and negative values, like
plane waves do.

In ODEON, different simulation methods are used in combination to ensure optimal performance
and precision. These are the Image Source Method (ISM), Early-Scattering Method (ESM), Ray-Tracing
(RT) and Ray-Radiosity Method (RRM). The first three methods collaborate in the early-reflections
part, while the last one is essentially the late-reflections part. In point source calculations both the

133
‘early’ and ‘late’ reflection parts are used. In line or surface source calculations, only the ‘late’
reflection part is used.

Transition order
Every reflection is classified by an order, according to how many walls have been involved in the
reflection path. In ODEON, a transition order is used to determine where the early-reflections part
stops and the late-reflections part begins. The default value is 2, which means that all reflections
that involve 2 walls are calculated according to the early-reflections part. Reflections of order
higher than 2 (more than 2 walls involved) are calculated according to the late-reflections part.

Energy

Early- reflections Late reflections

ESM

ISM
RRM

RT
Reflection order
Ray tracing used to
detect image sources Transition order

Note: All the three methods may overlap in time, since the transition order can affect reflections at
different times. The transition order criterion ensures that energy conservation is maintained during the
gradual shifting from one method to the other.

User inputs
Various parameters controlling the calculation methods are accessible by the user in the Room setup.
The Number of late rays and Transition order are of special importance, and should always be evaluated
by the user (see Chapter 10).

Early-reflections part
During the early-reflections part the Image Source Method (ISM) and Early-Scattering Method (ESM)
are used to determine the reflections and scattering originated from point sources up to the
transition order. These two methods provide high deterministic accuracy in tracking the early
reflections, which are probably the most important in the whole impulse response. The early
reflections towards the receiver provide significant information about the origin of sound and the
distance from the source.

Image Source Method

An image source is a mirroring of the source behind a wall. In the classical implementation, Image
Sources are found by just taking all combinations of walls up to a given reflection order, e.g. Source-
Wall1-Wall2, Wall1-Wall3 and so on. After that, a visibility check has to be performed to determine
which image source is actually visible from the receiver. If not, the image source is discarded. For
simple rooms with few surfaces construction and visibility check can be easily accomplished.
However, for rooms with thousands of surfaces, developing the whole tree of combinations can

134
lead to a huge number of image sources, with a significant percentage being not visible – therefore
being discarded afterwards.
In ODEON, a more efficient approach is utilized. A fast ray tracing algorithm is initiated from the
source in order to help ‘detecting’ only visible image sources. Therefore, only the actual amount of
valid image source is generated. The algorithm works as it follows:
• A number of early rays is generated from the
source (ODEON suggests a number based on
room dimensions, Transition order and surface
sizes).
• The rays are reflected specularly until they reach
the Transition order.
• For each reflecting surface, a new image source is
added to an image source tree. The number of
image sources stored after this process is usually
much lower than all combinations of reflecting
surfaces, but the images are all valid because
hidden surfaces and most very small surfaces are
excluded. The calculation so far is receiver-
independent.
• For each of the image sources, ODEON performs
a visibility check by following the corresponding Visible and invisible images. Images S1 and S21 are
reflection path back to the source. If this path is visible from R, while S2, S3 and S12 are not.
unobstructed until the source, then the image
source is found visible, and a reflection is added to the impulse response. However, the
reflection path may be blocked by other surfaces, or it may not hit the right surfaces, for instance
if they are too small. In such cases, the image is hidden and does not contribute to the impulse
response. The figure above illustrates the concept of visible and hidden image sources.
Specifically, we focus on three first-order image sources S1, S2 and S3, generated from source S,
behind walls 1, 2 and 3, respectively. More image sources can be generated behind the other
walls, but we only focus on these three at the moment. The image sources S1, and S2 generate
further second-order image sources S12 and S21 behind the extension of walls 2 and 1,
respectively. The line connecting S2 and receiver R does not pass through the corresponding
wall 2, therefore that reflection cannot exist. In this case S 2 is not visible from the receiver. The
same happens with S3 and S12. They are not visible from the receiver through walls 3 and 2,
respectively. At the end, only sources S1 and S21 are visible and are counted in the impulse
0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 metres
response.

Image Receiver
source
3

Source
P2
1

Source: 2
Surface: *Receiver*
Refl.: 1
Path <m>: 33.12
Odeon©1985-2004 Time <ms>: 96

135
Early Scattering Method
For each visible image source that is generated, a range of early secondary sources is generated on
the surface the image source corresponds to. These account for scattering in the following way:
• These early secondary sources will emit early scatter rays from the image source’s surface as
defined by the Number of early scatter rays parameter. The early scatter rays will be traced from the
current reflection order and up to the Transition order. At each reflection point of the early
0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 metres
scattering rays, including the point of origin, a new early secondary source is created.
• Every early secondary source visible from the receiver adds a reflection to the impulse response.

Receiver
Early
secondary
3
Image
sources
source
P2
1
Source

Source: 2
The attenuation of an image or early secondary source is calculated taking the following
Surface: *Receiver* into
Refl.: 1
account: Path <m>: 33.12

• Directivity
Odeon©1985-2004 Time <ms>: 96
factor of the primary source in the relevant direction of radiation.
• Absorption coefficients of the surfaces involved in generating the image (can be angle
dependent if Angular absorption is enabled in the Room setup).
• Air absorption due to the length of the reflection path.
• Scattering loss (frequency dependent), due to the scattered energy which is handled by a
‘scattering tree’. Based on the scattering of the surface, the sound energy is distributed between
image sources (specular contribution) and early secondary sources (Lambert sources) from the
same surface. If new image sources are again generated based on the current image source (if
the Transition order has not been reached yet), the distribution will follow a scattering tree because
of energy conservation principles. This allows a realistic calculation of early scattering. For more
information about scattering, read section 9.5.

Example
In short, each time ODEON detects an image source, an inner loop of (scatter) rays (not visualised
in the 3D Investigate Rays display) is started, taking care of the scattered sound which is reflected from
this image source/surface.
If all scattering coefficients in a room are 0.5, then the specular energy of a first order image source
is multiplied by (1-0.5) - and the specular energy of a second order image source is multiplied by
(1-0.5)*(1-0.5). The scattering rays handle the rest of the energy.

Late-reflections part
The Late-reflections part uses an ODEON-modified Ray-tracing method, called Ray-radiosity
method, which proves to be very efficient and receiver-interdepended. In the classical
implementation of Ray-tracing, rays are thrown from the source, get reflected and traced by the
algorithm. A detection sphere is formed around the receiver, so that whenever a ray intersects the
sphere, the corresponding the reflection is added at the impulse response. Two main disadvantages
of this method are that:

136
• the size of the detection sphere is quite arbitrary - despite of the various methods used to specify
it.
• the ray tracing needs to be re-run for every new receiver.

The Ray-radiosity method removes these disadvantages by dividing the process in two parts, the
Rendering and Gathering. The rendering part of the calculation only stores geometrical information
which does not involve material data nor receiver position. For the gathering part, the sources are
“collected” by the receiver adding information to the reflectogram.

Rendering (receiver-independent part)

The purpose of this process is to find virtual sources that radiate energy into the room. Rather than
thinking of reflections, the idea is to substitute the room with a number of sources at various
positions in space, each of the sources given the appropriate delay.
The process is divided into an early part (for point sources only) and a late part (for all source
types).
• A number of late rays specified in the Room setup are emitted from the source.
• These will hit the surfaces of the room, be reflected, hit new surfaces, be reflected again, and so
on until reaching the Max. Reflection order as specified in the Room setup. They will be reflected
according to the vector based scattering method (described later) and taking into account the
scattering due to the surface size and surface roughness.
• Each time a ray gets reflected; its reflection order is increased by one. The first orders of
reflections have been included already during the Early-reflections part; therefore, no action is
taken for the ray up to the Transition order set in the Room setup.
• Once the reflection order is higher than the Transition order, the ray will start generating one late
secondary source every time it hits a surface. This process creates a map of late secondary sources
all over the room and we call it rendering.
• 0.00
The late secondary
5.00
sources
10.00
may have
15.00
a Lambert , Lambert 25.00
20.00
Oblique or Uniform directivities, depending on
30.00 metres
the calculation settings.

P1

Each time a ‘late’ ray hits a surface, a late secondary source is generated. The process is
repeated for all the late rays set in the Room Setup.
Odeon©1985-2004

The late reflection process does not produce an exponential growing number of reflections
(with respect to time), as the image source method would suggest. Instead, it keeps the same
reflection density during the whole calculation, allowing for reasonable calculation times. The
attenuation of a secondary source is calculated taking the following into account:

137
 • Directivity factor of the primary source in the relevant direction of radiation (used with
point sources only).
• Absorption coefficients of the surfaces involved in generating the image (can be angle
dependent if Angular absorption is enabled in the Room setup).
• Air absorption due to the length of the reflection path.
• Directivity factor for secondary sources e.g. the Lambert, Oblique Lambert or Uniform directivity, see
later.

Remember: The rendering (receiver-independent part) stores a large file containing the ray history.
These files will speed up calculations of other receivers, making use of the same late secondary sources.
However, if these files are not erased when finishing a project, you may gradually fill up your hard drive.
Use File>Delete files>Delete calculation files to save space from such history files, which can be easily
recalculated.

When all calculations in a room have been carried out, use the File>Tidy directory tool to free some more
disk space. The Tidy directory tool will clean a complete directory and its sub-directories and delete all
calculation files for all rooms (A directory in this context can also be the C:\ drive in which case it will be
a lengthy process).

Gathering (receiver-dependent part)

In this part, the information from all generated late secondary sources is collected (gathered) to a
desired receiver location. ODEON checks each secondary source to determine whether it is visible
from the receiver. If it is visible, the contribution is added to the impulse response.
At this point the contributions of direct and reflected sound are collected at the receiving point
allowing the calculation of the results known as Single Point, Multi Point, Grid Response and 3DMatrix.
When more than one receiver is involved, the receiver-dependent part of the process is simply
repeated for each receiver. When more than one source is involved, the response at a given receiver
is simply0.00the superposition
5.00 of the responses
10.00 15.00 from
20.00 the individual
25.00 metres sources, each one delayed
appropriately, if a delay is applied to the source.

P1

After the ray tracing process is completed, the receiver will ‘collect’ the contribution to
the impulse response from the secondary sources.
Odeon©1985-2004

Remember: ODEON automatically takes care of handling which of the calculations and result files are
currently consistent with user-entered data, erasing those that are no longer valid. Thus, in some
situations you may experience that ray tracing calculations have already been done and are still valid,
in other cases they have to be recalculated.

138
One of the advantages of the late ray tracing method (ray radiosity) used in ODEON compared to
more traditional methods is that rays do not even have to come near to the receiver to make a
contribution - and a late ray, being reflected 100 times, has a potential of generating 100 secondary
sources. Thus, even in coupled rooms, with only a modest number of rays, it is possible to obtain
a reasonable number of reflections at a receiver (which is required to obtain a result that is
statistically reliable). This gives a fine balance between high reliability of the calculation results and
short calculation time.

A complete histogram, containing both early and late energy contributions, is generated and used
to derive Early Decay Time and Reverberation Time. The other room acoustical parameters are
calculated on basis of energy collected in time and angular intervals.
For surface sources and line sources, a number of secondary sources are placed randomly on the surface
(or line) of the source, each emitting one ray, and radiating a possible contribution to the receiver.
The rays emitted from these source types generate an independent secondary source each time
they are reflected. Compared to the calculation principle applied to the point sources, you might
say that Transition order is always zero, and that only late energy contributions are collected for these
source types, or rather that calculations are based exclusively on ray-radiosity. For point sources,
only the late reflections are handled like surface sources and line sources. For the first couple of
reflections, a method including image sources is used.

Summarizing the calculation method used for point response

calculations in ODEON
We can summarize the descriptions above in the following steps. The early-reflection part is
described by steps 1 to 7, while the late-reflection part by steps 8 to 11:
1. The early ray tracing is initiated from the source. Rays are reflected specularly.
2. Each time the ray collides with a surface, a new image source is added to the image source
tree. Since the image source results from an existing reflection, it is valid at the same time.
3. Soon all valid image sources have been generated behind each wall, up to the specified
transition order.
4. ODEON checks each image source to determine whether it is visible from the receiver.
5. If the image source is visible, then a reflection is added to the reflectogram and a range of
early secondary sources are generated on the surface the image source corresponds to.
6. ODEON simulates a surface source, which emits the number of early scatter rays up to the Transition
order.
7. Each ray that reaches the receiver without colliding with any obstacle, adds a reflection to the
reflectogram.
8. The late rays are emitted from the source and are reflected around the room, taking the
scattering of the surfaces into account. The rays start to be traced from the very first orders of
reflection, but no action is taken up to the Transition Order, since it has been taken into account
in the Early-reflections part already. The late rays are reflected until the Max. Reflection order.
9. Late secondary sources are generated at each reflection point (when a ray hits a surface).
10. The ray path is traced from the late secondary sources directly to the receiver.
11. ODEON checks each late secondary source to determine whether it is visible from the receiver.
If visible, the contribution is added to the reflectogram.

139
The following figure illustrates the process of adding reflections during the early and the late part.

A representation of the contributions added to the reflectogram.

The reflections generated by the image sources and early secondary sources, during the early
reflection method, occur before the Transition order. Generally, the reflections from early secondary
sources are reinforcing the reflection from the image source related to the same surface. After the
transition order, reflections are generated by the late secondary sources using the late reflection
method. In reality, all three methods will overlap in time as the transition criterion is the transition
order which is not strictly linked to time. This creates a smooth transition between the methods
creating no sudden jumps in the decay of sound.

9.4 Angle-dependent absorption

In reality sound absorption of materials depends on the angle of incidence. However, information
on angle dependent absorption is difficult to measure and rarely available. Therefore, ODEON has
implemented algorithms which derives angle dependent absorption from the random incidence
absorption coefficients which are commonly available – in the Odeon material library, from
manufacturers of acoustics materials etc.

Absorption is angle dependent, but will work differently between “soft” and “hard” materials.
Hard concrete would be an example of a hard material, and a porous absorber would be an
example of a soft material. ODEON determines if the material can be seen as a soft material, hard
material or something in between from the absorption coefficients.

The average absorption coefficient for the 1k, 2k and 4k Hz octave bands gives a value that is used
as follows:
• If the average value is exactly 0.3, the random incidence absorption coefficient for the material
(as set in the material list) will be valid for all angles of incidence.
• If the average value is exactly 0, the material is seen as a hard material and the absorption will
be affected by the angle of incidence as explained further below.
• If the average value is exactly 1, the material is seen as a soft material and the absorption will
be affected by the angle of incidence as explained further below.
• If the average is between 0 and 0.3, the material is something in between hard and not. The
absorption will be a mix of the random incidence absorption coefficient (as given in the material
list) and the angle absorption for hard materials. A material can be more or less hard, and
therefore the mix is based on linear interpolation from where it is placed between 0 and 0.3.

140
• If the average is between 1 and 0.3, the material is something in between soft and not, and the
absorption will be a mix of the random instance absorption coefficient (as given in the material
list) and the angle absorption for soft materials. A material can be more or less soft, and therefore
the mix is based on linear interpolation from where it is placed between 1 and 0.3.

The absorption for hard and soft materials will change based on the angle of incidence of the sound.
On one end we a have perpendicular, direct incidence, which is said to have an angle of incidence
of 0 degrees. On the other end we have a parallel, grazing, incidence which corresponding to an
angle of incidence of 90 degrees. And then we have that:

• Hard materials: the absorption will be higher the larger the angle of incidence (that is, the closer
the angle is to 90 degrees). And the angle will affect the higher frequencies the most. See graphs
below.
• Soft materials: it will be the opposite from hard materials. The absorption will be higher for
smaller angle of incidence, but the angle will still affect the higher frequencies the most.

The statistical information can be seen in the graphs below from (Rindel 1993).

Hard material (high impedance). Soft material (low impedance).

Calculated absorption coefficient as a function of angle of incidence and with the frequency as the parameter. A statistical absorption
2
coefficient of 0.5 and a square area of 11 m has been assumed. (Rindel 1993).

Hard material (high impedance). Soft material (low impedance).

Calculated absorption coefficient as a function of frequency and with the angle of incidence as the parameter. A statistical absorption
coefficient of 0.5 and a square area of 11 m2 has been assumed. (Rindel 1993).

141
9.5 Scattering
This chapter covers the calculation methods used for scattering. For a guide on how to choose a
correct scattering coefficient for a project/material, please look in the Materials chapter (chapter 4)
of this manual.

Scattering in ODEON uses multiple different methods to get the best results. This, however, can
make it a bit difficult to fully understand as a user, but for most users it is not necessary to
understand the details in order to use scattering.

The acoustical calculations in ODEON are divided into two parts: early and late. Scattering is dealt
with two different ways in the two parts. Besides this, scattering is both surface size dependent and
frequency dependent, and ODEON takes this into account.

A total scattering coefficient is applied to each surface based on both the user chosen coefficients
and, if active, calculations of extra scattering from the size of the surface and distance to the source.

Scattering is part of three different calculations. Very briefly, the following can be stated (explained
in more detail further below):

• The reflections of the rays used in ‘late’ rays will be based on the scattering coefficients.
This scattering is not influenced by frequency.
• The late secondary sources (Oblique Lambert Sources) made by ‘late’ reflections will be
based on the scattering coefficients, angle of incidence, and frequency of the octave band.
• The ‘early’ reflections will make normal Lambert Sources and Image Sources, and the
distribution of sound energy between them will be based on the scattering coefficients.

How scattering generally works for the late reflections:

• A chosen number of so-called ‘late rays’ is emitted from the sound source. When a ray hits
a surface, it will get a reflected direction. The calculation of the reflected direction is based
on the user-chosen scattering for the surface and if Reflection based scattering is enabled; on the
angle of incidence, the size of the surface and distance from the source. For every point
where a ray hits a surface, a late secondary source is generated.
• If a late secondary source is visible to a receiver, it will have influence on the decay curve and
the acoustical results. The amount of sound power is based on the absorption coefficients
the ray has met, but also (if active) on the Oblique Lambert method. The Oblique Lambert is based
on the scattering coefficient of the surface, the angle of incidence, and frequency of the
octave band.

How scattering generally works for the early reflections (explained in detail further below):

• For the early reflections, up to the set transition order, another method involving mirror
sources is used.
• Beside an image source, there will be made a number of early secondary sources on each of
the surfaces involved in the early reflection paths. The 0. order early secondary sources will
follow Lambert directivity. For higher reflection orders up to the transition order Oblique
Lambert is applied, in any the directivity will be different for each octave band, according
to the scattering coefficients.
• The sound power will be distributed between the image sources and the early secondary
sources according to the chosen scattering for the surface. If 100% scattering, only the early
secondary sources will be active, if 0% scattering only the image source.

142
Vector Based Scattering – reflecting a ‘Late ray’
Vector based scattering is an efficient way to include scattering in a ray-tracing algorithm. The
direction of a reflected ray is calculated by adding the specular vector according to the laws of
reflection, scaled by a factor (1 − s) to a scattered vector with random direction (where s is the
scattering coefficient), following the angular Lambert distribution of ideal scattered reflections:
Sin (2 ) (Rindel, 1995) which has been scaled by a factor s . If s is zero, the ray is reflected in the
specular direction, and if s equals 1, the ray is reflected in a completely random direction. Often
the resulting scattering coefficient may be in the range of say 5 to 20 percent, and in this case, rays
will be reflected in directions which differ just slightly from the specular one, but this is enough to
avoid artefacts due to simple geometrical reflection pattern.

Incident
Specular (weight: )

Resulting

Scattered (weight: s )

Vector based scattering. Reflecting a ray from a surface with a scattering coefficient of 50% results in a reflected
direction which is the geometrical average of the specular direction and a random (scattered) direction. Note:
Scattering is a 3D phenomenon, but here it is shown in 2D for simplicity.

The Reflection Based Scattering coefficient

When the reflection based scattering coefficient is activated in the Room setup, ODEON will do its
best in estimating the scattering introduced due to diffraction. The user defined scattering
coefficients assigned to the surfaces should only include scattering occurring due to surface
roughness, then diffraction phenomena are handled by ODEON. In many cases a scattering
coefficient of, say 5%, for all smooth surfaces may be sufficient. The Reflection Based Scattering
Method combines scattering caused by diffraction (due to typical surface dimensions, angle of
incidence, incident path length and edge diffraction) with surface scattering. Each of the two
scattering effects is modelled as frequency dependent functions. The benefits are two-fold:
• Separating the user defined surface scattering coefficient from the scattering that results room
geometry, makes it easier for the user to make good estimates of the coefficients.
• Scattering due to diffraction is distance and angle dependent, and as such it is not known before
the source and receiver are defined and the actual ‘ray-tracing’ or image source detection takes
place. One example could be a desktop providing a strong specular component to its user,
whereas it will provide scattered sound at remote distances.

The method has several advantages. For starters, it makes life easier for the user because the same
scattering coefficient can be used for different surfaces, no matter their size. It also allows a better
estimate of the actual scattering occurring at a reflection point, because scattering caused by
diffraction is not fully known before the actual reflections are calculated, thus angles of incidence,
path-lengths, etc., are known. In order to allow these features to be included in predictions, the
Reflection Based Scattering Coefficient sr is calculated individually for each reflection as
calculations take place, by combining the surface roughness scattering coefficient ss with the
scattering coefficient due to diffraction sd :
sr = 1 − (1 − sd )  (1 − ss )

143
Ss -Surface scattering
Assigning materials to the surfaces,
the scattering coefficient can be Frequency dependent scattering
defined by the user in the surface 1.00

Scattering coefficient
list. This is the scattering coefficient
0.80
at mid frequencies (logarithmic 0,90
average of 500 and 1000 Hz). 0.60 0,80
0.40 0,55
However, for the calculations,
ODEON uses scattering 0.20 0,25
coefficients that vary with the 0.00 0,05
frequency. The actual scattering
coefficients used in each octave 0,01
7band are shown in the figures. Frequency, Hz
Interpolation between the curves is
applied for other values. The scattering of a material changes due to frequency. The user chooses
the scattering for the mid frequency (707Hz). This value determines a
Of the two figures, one is shown curve, which is extrapolated towards high and low frequencies, and is
used for the frequency dependent calculations in ODEON.
with a linear y-axis and the other
with a logarithmic y-axis.
Scattering coefficients are best
displayed with a logarithmic axis.
Frequency dependent scattering
The scattering coefficient decreases 1.0000
Scattering coefficient

towards low frequencies, and

theoretically the slope is a factor of 0.1000 0,90
4 per octave. This is clearly shown 0,80
0.0100
as straight lines in the display with 0,55
logarithmic axis. 0.0010
0,25
Very small scattering coefficients 0.0001 0,05
4000
125
250
500
1000
2000

8000
63

can be applied for very smooth

0,01
surfaces, but the scattering
Frequency, Hz
coefficient can never be 0.
Same as above, but with a logarithmic y-axis. When changing scattering
coefficient, a doubling is comparable in a logarithmic axis. For example,
a change from 0.05 to 0.1 is comparable to a change from 0.1 to 0.2.

Remember: The scattering coefficients influence the simulations, and the effect of changing the
scattering coefficient depends on the relative change. For example, a change from 0.05 to 0.1 is
comparable to a change from 0.1 to 0.2. Scattering coefficients higher than 0.5 means in practice a
strong scattering, and a further increase will change very little.

Sd -Scattering due to diffraction

In order to estimate scattering due to diffraction, reflector theory is applied. The main theory is
presented in (Rindel, 1986; Rindel, 1992). The goal in these papers was to estimate the specular
contribution of a reflector with a limited area; given the basic dimensions of the surface, angle of
incidence, incident and reflected path-lengths. Given the fraction of the energy which is reflected
specularly we can however also describe the fraction s d which has been scattered due to
diffraction. A short summary of the method is as follows: For a panel with the dimensions l  w :
above the upper limiting frequency f w (defined by the short dimension of the panel) the frequency

144
response can be simplified to be flat, i.e. that of an infinitely large panel. Below f w the response
will fall off with by 3 dB per octave. Below the second limiting frequency f l , a decay of 3 dB per
octave is added, resulting in a decrease of 6 dB per octave. In the special case of a quadratic surface,
there will only be one limiting frequency below, which the specular component will decrease of 6
dB per octave. The attenuation factors K l and K w are estimates to the fraction of energy which is
reflected specularly. These factors take into account the incident and reflected path lengths (for ray-
tracing we have to assume that reflected equals incident path length), angle of incidence, and
distance of reflection point to the closest edge of the surface.

1 for f  fw 1 for f  f l
 
Kw =  f Kl =  f
f for f = f w f for f = f l
 w  l

ca* ca*
fw = , fl =
2( w  cos  ) 2 2  l2

Where:
d inc  d refl
a* =
2(d inc + d refl )

If we assume energy conservation, then we must also assume that the energy which is not reflected
specularly has been diffracted - scattered due to diffraction. This leads to the following formulae
for our scattering coefficient due to diffraction:

sd = 1 − K w K l  (1 − se )

In order to compensate for the extra diffraction which occurs when a reflection appears close to an
edge of a free surface, the specular component is reduced by a factor (1 − se ) . The edge scattering
coefficient is defined to be 0.5 if the reflection occurs at the edge of a surface saying that half of the
energy is scattered by the edge and the other half is reflected from the surface area. If the reflection
point is far from the edge, the edge scattering becomes zero – initial investigations suggests that
edge scattering can be assumed to zero when the distance to the edge is greater than approximately
one wave length, therefore we define the edge scattering coefficient as:

 c
0 for d edge  cos  
 f
se= 
0.5(1 − d edge  cos   f ) for d c
edge  cos  

 c f

As can be seen, scattering caused by diffraction is a function of a number of parameters, of which

some are not known before the actual calculation takes place. An example is that oblique angles of
incidence lead to increased scattering, whereas parallel walls lead to low scattering, and sometimes
flutter echoes. Another example is indicated by the characteristic distance a * , if source or receiver
is close to a surface, this surface may provide a specular reflection even if it is small, on the other
hand if far away it will only provide scattered sound, sd  1 .

145
 log(E)

log(frequency)

Energy reflected from a free suspended surface given the dimensions 𝑙 ∙ 𝑤. At high frequencies the surface
reflects energy specularly (red), at low frequencies, energy is assumed to be scattered (blue). 𝑓𝑤 is the upper
specular-cut off frequency defined by the shortest dimension of the surface, 𝑓𝑙 is the lower cutoff frequency
which is defined by the length of the surface.

Boundary walls and interior margin

As long as surfaces are truly freely suspended surfaces, they will act as effective diffusers down to
infinitely low frequencies. For surfaces which are elements in the boundary of the room (such as
windows, doors, paintings, blackboards etc), it is not provided effective scattering down to
infinitely low frequencies. From diffuser theory, it is found that the typical behaviour is that the
effectiveness of a diffuser decreases rapidly below a cut-off frequency, which can roughly be
defined from the depth of the diffuser (wall construction) being less than half a wavelength. At two
octave bands below the cut-off frequency, the diffuser is no longer effective. At the lowest
frequencies however, the dimensions of the room will provide some diffraction, therefore the
dimensions of the reflecting panel as used in the formulae for fl and f w apply for the lowest
frequencies, substituted with the approximate dimensions of the room. For frequencies in-between
high and low, a combination of surface and room dimensions are used. It is worth noticing that it
is not only the depth of the wall construction which enables the elements of the wall construction
to provide diffraction. The angling between the surfaces, offsets e.g. the door being mounted in a
door hole, or the surfaces being made of different materials, all provide phase shifts which result
in diffraction. Therefore, it may be reasonable to assume that the boundary walls have a minimum
depth of, say, 10 cm, in order to account for such phase shifts.

Boundary Furniture

Interior Margin

Ground plan of a room with the interior margin marked as a dashed green line. Surfaces between the boundary
and the interior margin (black) provide low scattering at low frequencies, while surfaces inside the interior
margin (green) provide high scattering throughout the whole frequency range.

146
The typical depth of the surfaces attached to the wall such as windows and blackboards, should be
specified in the Interior margin in the Room setup. ODEON will use this number in order to distinguish
between interior and boundary surfaces. In case any surface is inside this margin, it will be
considered as a boundary surface. Once the margin has been entered and the Room setup dialog has
been closed the 3DView will display surfaces which are considered to be interior in a dark green
colour, while the boundary is displayed in black. Diffraction from the boundary will be calculated
taking into account that diffraction is limited towards the lowest frequencies because of limited
depth of the wall constructions.

Limitations
In special cases, the Reflection Based Scattering Coefficient may overestimate the scattering provided
by small surfaces which are only fractions of a bigger whole, e.g. small surfaces being part of a
curved wall or dome. Such surfaces should not cause diffraction due to their individual area,
because the individual surfaces do not provide any significant edge diffraction. In these cases, the
overestimation can be bypassed by setting the surface Type to Fractional in the Materials List, see Chapter
4. When setting the Type to Fractional, the surface area used for calculating the Reflection Based
Scattering Coefficient is determined from the box subscribing the room, rather than the individual
surface - if the construction part of which the fractional surface is part of, is considerable smaller
than the ‘room box’ scattering, it might be underestimated and a higher scattering coefficient
should be assigned to the surface.

A typical diffuser depth (interior margin) entered by user, determines which parts of the geometry should be
considered exterior of a room, thus limiting the scattering of the boundary surfaces which cannot be
considered freely suspended at the lowest frequencies.

9.6 Oblique Lambert

In the ray-tracing process, a number of secondary sources are generated at the collision points
between walls and the rays traced. It has not been investigated yet which directivity to assign to
these sources. A straight forward solution, which is the one ODEON has been applying until
version 8 is, to assign Lambert directivity patterns. That is, if the diffuse area radiation is modelled
via cosine directivity. The result would be that the last reflection from the secondary sources to the
actual receiver point is handled with 100% scattering – regardless of the actual scattering properties
for the reflection. This is not an optimal solution; in fact, when it comes to the last reflection path
from wall to receiver, we know that not only the incident path length to the wall, but also the path
length from the wall to the receiver are available, allowing a better estimate of the characteristic
distance 𝑎∗ as in the case of the ray-tracing process, where 𝑑𝑟𝑒𝑓𝑙 was assumed to be equal to 𝑑𝑖𝑛𝑐 .
So which directivity to assign to the secondary sources? ODEON proposes a directivity pattern
which we call Oblique Lambert. Reusing the concept of Vector Based Scattering, an orientation of our
Lambert sources can be obtained taking the Reflection Based Scattering coefficient into account. If
scattering is zero then the orientation of the Oblique Lambert source is found by the laws of
reflection. If the scattering coefficient is equal to 1, then the orientation is that of the traditional
Lambert source, and for any scattering value between 0 and 1, the orientation is determined by the
vector found using the Vector Based Scattering method.

147
 Traditional Lambert directivity at the left and Oblique Lambert at the right. Oblique Lambert produces a
shadow zone where no sound is reflected. The shadow zone is small if scattering is high, or if the incident
direction is perpendicular to the wall. On the other hand, if scattering is low and the incident direction is
oblique, then the shadow zone becomes large.

If Oblique Lambert was implemented as described without any further steps, this would lead to
an energy loss because part of the Lambert balloon is radiating energy out of the room (notice the
incomplete circle in the second figure above). In order to compensate for this, the directivity pattern
has to be scaled with a factor which accounts for the lost energy. If the angle is 0, then the correction
factor is 1 as no correction is necessary, and if the angle is 90° the factor becomes 2, because half of
the balloon is outside the room. This is the maximum possible correction factor, and factors for
angles between 0° and 90° have been found using numerical integration.

Oblique Lambert attenuation factors

2,5
Correction coefficient

1,5

0,5

0
0 10 20 30 40 50 60 70 80 90
Oblique angle

Correction factor for Oblique Lambert. When the oblique angle is zero, Oblique Lambert corresponds to
traditional Lambert and the correction coefficient is 1. When the oblique angle is 90° corresponding to grazing
incidence on a smooth surface, the correction factor reaches its maximum of 2.

A last remark on Oblique Lambert is that it can include frequency depending scattering at (close
to) no computational cost. This part of the algorithm does not involve any ray-tracing which tends
to be the heavy computational part in room acoustics prediction. Only the orientation of the
Oblique Lambert source has to be recalculated for each frequency of interest in order to model
scattering as a function of frequency.

148
9.7 Diffraction over screens and round objects (Screen
diffraction)
When no direct sound is received from a point source at a receiver position, ODEON will try to
detect a one- or two-point diffraction path, using the methods suggested by Pierce in “Diffraction
of sound around corners and over wide barriers” (Pierce, 1974). By only calculating the diffracted
contribution when the point source is not visible, the contribution can be added to the impulse
response without considering the phase-interaction between direct sound and the diffracted
component, just like the reflections are added to the impulse response. Only diffraction from one
edge of the diffracting object is taken into account, the one with the shortest path-length. The edge
of diffraction is considered infinitely long. More information and validation is found in (Rindel, et
al., 2009).

By using the algorithms by Pierce, rather than simpler ones, ODEON is capable of handling more
complicated objects than the single surface screen, e.g. objects such as book shelves, the corner in
e.g. a L-shaped room and buildings out-doors or diffraction over a balcony front edge, or over the
edge in an orchestra pit.

Estimating diffraction points around objects

In order to calculate diffraction, it is essential that the diffraction path is established. In rooms with
moderate complexity, it becomes virtually impossible for the user to describe which surface edges
will act as diffracting edge, and therefore ODEON has a built-in detection algorithm. In order to
limit the complexity and time consumption of the detection, there are some limitations to the type
diffraction paths ODEON can detect. In order to understand the limits, here is a short description.

When a source is not visible from the receiver, ODEON will try to detect the shortest diffraction
path from source to receiver in the following way:
1. A ray is sent from source towards the receiver and the first surface hit is registered (A)
2. A ray is sent from receiver towards the source and the first surface hit is registered (B)
A diffraction path should follow the path Source-An-Bm-Receiver, where An and Bm are the nth and
mth point on an edge of each of the surfaces. The shortest path with full visibility between the three
sub paths; Source-An, An-Bm and Bm-Receiver is used for the further calculations.

In cases where there are other surfaces which block the paths between one of the sub paths,
ODEON will not detect a diffraction path.

An An Bm An Bm An Bm

R
R R R S
S S S

a) b) c) d)
Diffraction paths and their detection. a) Odeon detects a 1 point diffraction path over a thin screen. b) Odeon
detects a 2 point diffraction path over a wide barrier. C) Odeon fails to detect a diffraction path because the
path from An-Bm is obscured by a third surface – in fact it’s a 3 point diffraction path. d) Odeon detects a
diffraction path over two thin barriers.

149
Contribution from diffracted sound over a screen, around a corner or around a book shelf, is not
likely to change results noticeably. However, in rooms such as an open plan office with very
absorbing ceiling or in outdoor situations, diffraction around edges may play an important role.

9.8 Radiated rays from a source

In ODEON (Combined and Industrial version only) three different kind of sources are available;
the point source, the line source, and the surface source.

Point Sources
For Single, Multi and Grid response calculations (and for the 3D Investigate Rays and 3D Billiard displays) rays
are radiated in directions distributed as evenly as possible over a solid angle. To obtain the even
distribution, rays are emitted in directions given by the Fibonacci spiral. Using a spiral also ensures
that there will be no preference towards (or away from) e.g. horizontal rays as might be the case if
rays were radiated from a number of e.g. horizontal rings. For Quick Estimate and Global Estimate, the
radiated directions are chosen randomly, allowing the calculation to be finished after any ray,
without getting a very uneven distribution of radiated directions.

Surface Sources and Line sources (Industrial and Combined versions

only)
For these source types, points of origin for the rays and their initial directions are the same, no
matter the calculation type. For each radiated ray, a random point or origin point is chosen at the
line or surface source. From this point the ray is sent out in a random direction, and two types of
radiation can be chosen: Lambert and Spherical. When Lambert is chosen, the strength of each ray is
proportional to the cosine of the angle between its direction and the normal. Practically, rays
radiated parallel to the surface have zero strength, whereas rays perpendicular to the surface will
carry the maximum of the energy. Spherical radiation introduces uniform strength to the rays,
regardless of their direction.

Lambert radiation is preferable for ordinary surface sources, such as noise emission from
machinery. Spherical radiation is recommended when several point sources are modelled as a
surface source, instead of modelling each one individually (e.g. many people distributed over a
large canteen hall). Spherical radiation is also a good option for modelling traffic noise, which can
be considered as a dense array of vehicles, acting as individual point sources.

Lambert (left) and spherical (right) types of radiation, applicable to line and surfaces sources.

150
9.9 Processing reflection data for auralisation in Single
Point Response
A typical point response calculation in ODEON includes from 100.000 to 1.000.000 reflections per
source/receiver. The reflections are calculated in terms of time of arrival, strength in 8 octave bands
and angle of incidence (azimuth and elevation). The information on size of the reflecting surfaces
and absorption coefficients are also available as a part of the calculation.

Binaural filters for headphone playback: When the Auralisation setup|Create binaural impulse response file is
turned on, reflections are processed in order to create a binaural impulse response (BRIR). First of
all it is determined whether a phase shift should be applied to the reflection, based on surface size
and absorption coefficients of the last reflecting surface (Rindel, 1993). The reflection is
filtered/convolved through 9 octave band filters (Kaiser-Bessel filters, the ninth being extrapolated)
and finally the reflection is filtered/convolved through two corresponding directional filters, one
for each ear (Head Related Transfer Functions), creating a binaural impulse response for that
reflection. This process is carried out for each reflection received at the receiver point,
superimposing all the reflections as a result; the Binaural Room Impulse Response (BRIR) for that
particular receiver point is obtained. The actual order in which the filtering is carried out in
ODEON, differs from the description above (otherwise the calculation time would be astronomic),
but the BRIR results contain the full filtering with respect to octave band filtering in nine bands, as
well as directional filtering.

Calculation of B-format filters for external decoding and Surround filters for loudspeaker playback
is based on the Ambisonics technique which is covered in (Gerzon, 1992; Bamford, 1995; Malham,
2005; Furse). Most of the specific steps in involved in the generation of these filters are similar to
those used for generation of BRIR’s. HRTF’s are not used in the calculation, since it aims at
loudspeaker representation where the listener will receive reflections from the head and torso.

9.10 Calculation method for Reflector Coverage

25.000 rays are sent out from the selected source, and if the rays hit one of the surfaces defined as
reflector surfaces at the Define reflector surfaces menu, a cross is painted where the reflected rays hit
the room surfaces. Note that the value of the Transition order is taken into account; if it is zero and the
Lambert scattering is active, the chosen reflectors will show a degree of scattered reflection
corresponding to their scattering coefficients. Sound from line and surface sources will include
scattering upon reflections, if the Lambert scattering is on.

151
 Calculation Parameters
The calculation parameters specify how calculations are carried out in ODEON. As a starting point,
ODEON suggests values that may be considered suitable for most calculation parameters, but
please be aware that they may not apply perfectly to every situation, e.g. in a room with strong
decoupling effects or very uneven distribution of absorption more late rays will be needed
(explained in more detail further below), and in a room with strong focusing a transition order
higher than 2 is suggested. Select one of the pre-sets: Survey, Engineering or Precision, which represent
different levels of quality of results. ODEON will adjust the calculation parameters accordingly.
You can always adjust the calculation parameters further manually. There are a couple of
parameters in the General settings which should always be adjusted the Impulse response length and
Number of late rays. The Specialist settings normally don’t need to be adjusted.

There is more information on how to setup calculation parameters in the help file. Find it in
ODEON by pressing F1 when the Room setup window is open.

152
10.1 Impulse response length (Job calculations and
Global Estimate)
The Impulse response length should always be specified. It should be at least 2/3 of the longest
reverberation time over all 1/1 octaves. It determines how many milliseconds of “decay curve” are
to be calculated. It is a key parameter - if it is shorter than approx. 2/3 of the reverberation time in
the room, then T30 cannot be calculated (because the dynamic range of the decay curve is less than
35 dB), instead results for T30 will be displayed as *.** in result displays. For reliable results, it is
recommended to use an Impulse response length which is comparable to the estimated reverberation
time. To get an estimate of the reverberation time in a room, use the Quick Estimate available in
the Materials list. The maximum allowed Impulse response length is 200000 ms. It is tempting just
to use a very long impulse response length in order to avoid that reverberation time is not
calculated at some frequencies – however using a length many time the expected T30 may harm
quality of results because the very late and unreliable decay at say -200 dB will be included in the
regression line from which T30 and other reverberation parameters are derived.

10.2 Number of late rays (Job calculations)

The Number of late rays influences the density of reflections in the late part of the decay. Having too
few late rays is a very typical cause of errors, but on the other hand more late rays will extend the
calculation time.

In the Room Setup, ODEON can suggest a number of rays for a survey, engineering or precision calculation.
The number is derived taking into account the aspect ratio of the room as well as the size and
number of surfaces in the geometry. In short, this means that ODEON will suggest more rays for
very long room with many surfaces, than for a basically cubic room with few surfaces.
The suggested number of rays will be sufficient for many rooms, however in some cases more rays
may be needed in order to obtain good results (please read further down for an explanation).

The ODEON algorithms are hybrid methods based on the Image Source Method and Ray-Radiosity
for the early part of the energy, and Ray-Radiosity for the later part of energy. Therefore, there is
also a Number of early rays parameter, which will be described later.

More rays needed?

There is no definite way of telling if more rays are needed for a specific calculation, but to get an
idea there are some good tips.

A good thing to investigate is the decay curves. For all editions, a decay curve can be derived in
the Global Estimate in the top menu. And for ODEON Auditorium and Combined, a decay curve is
always calculated in the Single point response results.

If a decay curve has suspicious spikes/drops you may try to increase the number of late rays. Please
see the graph below. This can also be caused by other effects, and be a part of the actual acoustics,
but for most cases it is a sign of too few rays used.

153
It is also worth checking if the Global Estimate stabilizes slowly or the global decay demonstrates a
‘hanging curve’ effect. A rule of thumb is to let the Global Estimate run until the decay curve seems
stable, then use, say, between 0.1-1 times the number of rays used in the Global estimate to specify
the number of rays to be used in the calculation of the point responses (specified in the Room setup).

It is a good idea to experiment with different Number of late rays and study the decay curves in Single
Point Responses. If many surfaces are added to a model, in-between calculation, it is recommended to
re-specify the Number of late rays too.

ODEON also includes a tool to investigate the actual number of reflections per millisecond used in
the calculation of a point response. The tool is called Reflection density and it is included in the
calculated parameters of the Single point, Multi point and Grid response results. The Reflection_density
parameter is included and set to Visible in the Room Acoustics Parameter List.

Additional to the numerical value, a graph like the one shown below is displayed in the Single Point
response in the job list.

For reliable results, the reflection density parameter (Reflection_density) should be higher than 25
reflections/ms, as explained further below.

What could affect the numbers of rays needed

If the room model contains strong decoupling effects or uneven distribution of the absorption, it is
recommended to increase the Number of late rays, even if the calculated reflection density seems high
enough. Other complex geometry could also increase the needed number of late rays.

154
1) Strong decoupling effects: If a dry room is coupled to a reverberant room, then more rays may
be needed in order to better estimate the coupling effect. An example could be a foyer or a
corridor coupled to a classroom. If the room where the receiver is located is only coupled to the
room where the source is located through a small opening, then more rays are also needed.

2) Very uneven distribution of the absorption in the room: In some rooms the reverberant field in
the x, y and z dimensions may be very different. An example of this could be a room where all
absorption is located on the ceiling while all other surfaces are hard. Another example could be
an open-air theatre (modelled as having a ceiling with 100% absorption). In particular, if
surfaces are all orthogonal while having different materials in the x, y and z dimensions of the
room and if low scattering properties on the surfaces are used, then more rays should be used.

On the other hand, if multiple sources with similar source power are visible from all/or most
receivers, then the Number of late rays can be decreased significantly.

10.3 Specialist settings

Below the General settings are the Specialist settings, which should only be adjusted if you are an ODEON
expert or experimenting. One exception is if you are importing rooms from much earlier versions
of ODEON, then it will be a good idea to enable Screen diffraction to make sure diffraction around
screens is included in future calculations.

Impulse response details

Max. reflection order (Job calculations only)
is a stop criterion which determines how many times a ray can be reflected. Under
Max. reflection order
normal conditions it should be as large as possible. In that case the Impulse response length will be the
actual stop criterion and the Max. reflection order is only taken into account when stopping rays that
have been trapped between two very narrow surfaces. This number may be decreased if you are
only interested in the very early reflections. The Maximum allowed reflection order is 10000.

Impulse response resolution (Job calculations and Global Estimate)

The Impulse response resolution is the width of the time steps in the Impulse response histogram
in which the energy of the reflections is collected during a point response calculation. The
histogram is used for calculation of reverberation parameters such as EDT and T30. A resolution of
1 ms is recommended – in Odeon Auditorium and Combined a resolution of 1 ms allows special
scale bars for Music, Studio and Speech to be selected in the 3D part of the Decay curves and the Reflectogram
in the Single Point response (resolutions of 0.5, 0.25 and 0.1 ms will also allow that).

Early reflections:
Transition Order (Job calculations only)
The Transition order (TO) determines at which reflection order ODEON changes from the early Image
Source Method to the late Ray-Radiosity method.

In special cases you may want to alter the parameters to conduct special investigations. E.g. to
investigate a high order echo problem, you may want to increase the Transition order to get the
reflections displayed in the Single point response|reflectogram display (Auditorium and Combined
editions only). However, this will rarely result in better calculation results for the room acoustical
parameters. In such cases create an extra copy of the room using the Files|Copy files menu and conduct
the special investigations on the copy.

155
The Transition order applies only to point sources. The current recommendation is TO = 2 for most
rooms. For rooms that are heavily packed with various fittings or rooms where no or few image
sources are visible from the receivers, a lower Transition order of 0 or 1 may be used. If a Transition order
of zero is selected then point responses will be calculated using Ray Tracing/Ray-Radiosity only -
which has better robustness to cases where the wave front is cut into small pieces e.g. by apertures
in the room - a situation which is not handled elegantly by the Image Source Method.

Number of early rays (Job calculations only)

The Number of early rays is normally best left for ODEON to decide. ODEON will estimate a Number
of early rays to use in order to detect important early reflections (Image sources) based on the number
and size of surfaces, the dimensions of the room as well as the user defined Transition order. It is not
given that more is better.

Number of early scatter rays (Job calculations only)

If the specified Number of early scatter rays is greater than 0, the early reflections from point sources are
split into a specular part (an image source) and a scattered part (given by the scattering
coefficients). 100 scatter rays (per image source) should be sufficient. The scatter rays are traced
from the reflection order of the images source up to the selected Transition order.

Select calculation methods

Angular absorption (Job calculations and Global Estimate)
When this option is on, angle dependent reflection factors are taken into account. The calculation
method uses the diffuse field absorption coefficients as input but generate angle dependent ones
taking into account angle of incidence and the size of the surface. The method is based on the work
in (Rindel, 1993). In some situations, this option does not improve the results (but on the other hand
it does not seem to harm) and in many situations it means a significant improvement. Three options
are available; Disabled, All materials and Soft materials only. The All materials is the thorough method whereas
the Soft materials only is a trade-off between calculation time versus prediction quality. The Soft materials
only method is usually almost as fast as the Disabled option and yields a quality almost as good as the
All Materials method which is somewhat slower.

Screen diffraction
If this option is checked, ODEON will try to calculate the
sound diffracted around objects when the source is not
visible from the receiver. The diffraction method is the
one suggested by (Pierce, 1974). This method includes
one and two-point diffraction, allowing diffraction over
or around single surface screens, wedge shaped screen
and two-sided objects. The Screen diffraction is only
carried out for the primary source (not for its image
sources) when it is not visible from the receiver. Screen
edges are detected automatically by ODEON and the
shortest path around an object is used for the
calculations. For more information on methods and
limitations please see Chapter 3. If one or more diffraction contributions are included in a point
response, they can be viewed as part of the Single point response>3DReflection paths when all reflections
are displayed (A-shortcut).

Surface scattering (Job calculations, Global Estimate and Quick Estimate)

If Surface scattering is set to Actual, all directions of late reflections are calculated using the scattering
coefficients assigned to the surfaces in the Materials list or according to the reflection based scattering

156
method (see below). If the scattering coefficient is 10%, then the vector of new ray direction will be
calculated as 90% of the specular vector plus 10% of the scattered vector (random direction due to
the Lambert distribution). If the Scattering method is set to None (s = 0), scattering is not taken into
account, thus all reflections are calculated as specular. If the surface scattering is set to Full scatter (s
= 1) all surfaces will scatter 100%. The Full and None scattering methods are NOT recommended
except for educational or research purposes.

Oblique Lambert
The Oblique Lambert method allows frequency dependent scattering to be included in late
reflections of point response calculations - this option is recommended.

Reflection Based scattering

The Reflection based scattering method automatically takes into account scattering occurring due to
geometrical properties such as surface size, path lengths and angle of incidence. The use of this
method is recommended unless that part of scattering has already been included in the scattering
coefficients assigned to the room’s surfaces.

Interior margin
By default, the typical geometrical offset in the boundary of the room is 10 centimetres. Surfaces
which are closer to the boundary surfaces of the room than the distance specified by the Interior
margin will also be considered boundary surfaces - this means that surfaces such as doors and
windows which may be modelled as being slightly on the inside of the boundary walls, will still
be considered as boundary surfaces. Interior surfaces are displayed in a green colour (teal) in the
3DView whereas boundary surfaces are black, so a change to the Interior margin will be reflected in this
display when the Room setup dialog is closed. This measure tells ODEON that the effective scattering
provided by boundary room surfaces should be restricted below a frequency derived from this
measure. To get an idea of our suggestions to this value please look into the geometries supplied
with the installation of ODEON - whether a value of 10 or 20 centimetres is chosen may not be
critical, but for rooms with a very 'jumpy' boundary it should be considered to specify this
parameter.

Key diffraction frequency

In order to obtain the best result in the mid-frequency range for speech and music, the default
diffraction frequency is 707 Hz. This is the frequency at which diffraction is calculated for the ray-
tracing part of calculations. All other parts of point response calculations take into account
frequency dependent scattering. This frequency should only be changed in special cases where the
focus is on another frequency range.

Scatter coefficients > x.xx handled as uniform scatter

From published material on measured scattering coefficients, there seems to be a general trend that
modest scattering tends to be area based whereas high scattering is better represented by uniform
scatter. In ODEON, small scattering coefficients below a certain value referred to as “x.xx”, is
handled either with the Lambert Oblique or Lambert algorithms, whereas scattering coefficients
above x.xx are handled using uniform scattering. Scattering coefficients in the context above are
the Reflection Based Scattering coefficients (if that option is activated); a typical result of this algorithm is
that reflections close to an edge will be handled with uniform scattering which is desirable.
Through empirical studies we have found that x.xx = 0.5 yields best results. If using a value of 0,
then all reflections are handled uniformly, and if using a value of 1, then all reflections are handled
using Lambert or Lambert Oblique as was the case in ODEON 8.0 and earlier.

157
 Achieving Good Results
The following chapter discusses how to obtain good results and indeed what is a good result. This
is not a straight answer on how to obtain the best result, but merely a discussion that may provide
some ideas as to what can be done in order to obtain reliable results in a program such as ODEON.

11.1 What are Good Results

The desirable precision - subjective limen
Before discussing how to achieve good results, it is a good idea to outline just what a good result
is. The subjective limen (or “just noticeable difference” - JND) on room acoustical parameters
should give a good suggestion as to the desirable precision. If the error between the ‘real’
(measured with some precision) and the simulated room acoustical parameter is less the one
subjective limen, then there is no perceivable difference and there would be no need for more
precise results. In some cases, it will be difficult or even impossible to obtain results at this
precision, but in most of the cases, being slightly above 1 JND will still be satisfactory.

Parameter Definition Subj. limen

(ISO 3382-1, 2009) and (IEC 60268-16, 2011) for STI
T30 (s) Reverberation time, derived from –5 to – 35 dB of the decay curve 5 % rel.
EDT (s) Early decay time, derived from 0 to – 10 dB of the decay curve 5 % rel.
D50 (%) Deutlichkeit (definition), early (0 - 50 ms) to total energy ratio 5 % abs.
C80 (dB) Clarity, early (0 – 80 ms) to late (80 ms – ∞) energy ratio 1 dB abs.
Ts [ms] Centre time, time of first moment of impulse response or gravity 10 ms abs.
time
G (dB) Sound level related to omni-directional free field radiation at 10 1 dB abs.
m distance
LF (%) Early lateral (5 – 80 ms) energy ratio, cos2(lateral angle) 5 % abs.
STI (RASTI) Speech Transmission Index 0.03 abs.
Room acoustical parameters and their subjective limen as given by Bork (Bork, 2000) and Bradley
(Bradley, 1986).

Example 1:
If the real G value is 1 dB and the simulated is 1.9 dB then the difference is not noticeable.

Example 2:
If the real LF value is 12 % and the simulated is 17 % the difference is only barely noticeable.

Note! When comparing measured parameters to the ones simulated it should be kept in mind that
the measured parameters are not necessarily the grand truth as there are also uncertainties on the
measured result. These errors are due to limited tolerances in the measuring equipment as well a
limited precision in the algorithms used for deriving the parameters from the measured impulse
response (or similar errors if results are not based on an impulse response measuring method).
There may also be errors due to imprecise source and receiver positions.

158
11.2 Sources of Error
There are many sources of error in a room acoustical simulation, leading to results which are less
than perfect (within one subjective limen). Sometimes this is quite acceptable because we are just
interested in rough results, at other times we are interested in results as good as possible. In any
case, being aware of the sources of error may help getting the best possible results out of ODEON.
The sources of error (or at least some of them) are:

• The approximations made in the ODEON calculation algorithms.

• Inappropriate calculation parameters (e.g. number of late rays).
• Imprecise material absorption coefficients.
• Imprecise material scattering coefficients.
• Not accurate or optimal geometry definition for use in ODEON (e.g. too detailed).
• Not accurate measured reference data to which simulations are compared.

Approximations made by ODEON

It should be kept in mind that algorithms used by software such as ODEON are only a rough
representation of the real world. In particular, the effect of wave phenomena is included in the
calculations only to a limited extent. There is very little to do with this fact for you, the user, except
to remember that very small rooms and rooms with very small surfaces are not necessarily
simulated at high precision.

Optimum calculation parameters

A number of calculation parameters can be specified in ODEON. These settings may reflect
expected reverberation time, a particular shape of the room or a trade-off between calculation
speed and accuracy.

Number of Late rays

Too few rays used in simulations is a very common reason for errors. Please read about how to
find a reasonable number of rays in Chapter 10.2.

A short explanation:
ODEON by default specifies a suggested Number of late rays to be used in point response
calculations. For some rooms however, this number of rays will be too low. The rooms
could be coupled rooms, or rooms with very uneven absorption. A good way to test if the
number of rays is sufficient, is by inspecting the decay curves as shown in section 10.2.

Transition order
The Transition order is less often a source of error, but in very detailed models it can be. The Transition
order applies only to point sources. Current recommendation is TO = 2 for most rooms. For rooms
with a high level of geometrical detail, with many small surfaces or rooms where no or few image
sources are visible from the receivers, a lower Transition order of 0 or 1 may be used. Please read about
Transition order in Chapter 10.3.

Materials/absorption data
Wrong or imprecise absorption data is probably one of the most common sources of error in room
acoustic simulations. This may be due to lack of precision in the measurements of the absorption
data, limitations in the measurement method itself, or because the construction material in the
model was assumed based on guesswork – in any case it is a good idea to remember this and to
estimate the size of error on the material data as well as the impact on the simulated results. It has
been observed that absorption coefficients outside the range 0.05 to 0.9 should be used with care

159
(Christensen, et al., 2008). In the Material list in Odeon, there is a button which will limit the range of
absorption coefficients assigned to surfaces in a room to a selected range.

Solution if materials’ data is uncertain:

There is really not much to do about the uncertainty of material data if the room does not exist,
except taking the uncertainty of the materials into account in the design phase. If the room does
indeed exist and is being modelled in order to evaluate different possible changes it may be a good
idea to tweak uncertain materials until the simulated room acoustical parameters fits the measured
ones as good as possible.

In ODEON 13 the Material Optimization Tool based on genetic algorithms was introduced. An initial
estimate of materials is required. Then the algorithm searches for the best set of absorption
coefficients, by a genetic evolution process, which makes simulations match better with
measurements. See Chapter 13.

Absorption properties in a material library are often assumed by users to be without errors. This is
far from being the truth. For high absorption coefficients and at high frequencies the values are
probably quite reliable. However, low and even mid-frequency frequency absorption data and
absorption data for hard materials will often have a lack of precision.

Low frequency absorption

At low frequencies the absorption coefficients measured in a reverberation chamber are with
limited precision because:
• There are very few modes available in a reverberation chamber at lowest frequency bands.
• Low frequency absorption occurs partly due to the construction itself rather than its visible
surface structure. Often it may not be possible to reconstruct a complete building construction
in a reverberation chamber, and if reconstructing only a fraction of the wall in the reverberation
chamber it will have different absorption properties, because it becomes more or less stiff.

Hard materials
Hard materials such as concrete are often listed as being 1% or 2% absorbing. It may sound like a
difference of 0.5% or 1% is not a significant difference. However, if a room is dominated by this
material (or one of the dimensions of the room is), a change from 1% to 2% is a change of twice as
much absorption!

Materials/scattering coefficients
The knowledge on scattering coefficients is currently rather limited. Hopefully in the future, the
scattering coefficients will be available for some materials. Meanwhile the best that can be done is
to make some good guesses on the size of the scattering coefficients and to do some estimates on
the effect of uncertainty. Please read Chapter 4.

Measurements
The reference data which you may use to compare simulated room acoustical parameters against,
is not guaranteed to be perfect. We must accept some tolerances on the precision of the measured
parameters.

Receiver position(s)
Common errors are:
• to base the room acoustic design on simulations in one or only few receiver positions.
• to place the receiver close to a surface.
• to use too short source-receiver distance.

160
Source-Receiver distance
Point response calculations made in ODEON are to be compared with point response
measurements and as such the (ISO 3382-1, 2009) standard should be followed:

To obtain good estimates of reverberation time, the minimum source-receiver distance should be
used in order to avoid strong influence from the direct sound. The minimum source–receiver
distance according to ISO 3382-1 is:

V
d min = 2 (m)
c *T
where :
V is the volume of the room (m3).
c is the speed of sound (m/s).
T is an estimate of the expected reverberation time (s).

Thus, for a typical concert hall a source-receiver distance less than 10 metres should be avoided in
order to get good predictions (measurements) of the reverberation time.

Minimum distance from the receiver to the closest surface

If a receiver is placed very close to a surface, then results will be sensitive to the actual position of
the secondary sources generated by ODEON’s late ray method. If such a secondary source happens
to be very close to the receiver, e.g. 1 to 10 centimetres, this may produce a spurious spike on the
decay curve, resulting in unreliable predictions of the reverberation time – indeed if the distance is
zero then in principle an infinitely large contribution would be generated. To avoid this problem,
it is recommended that distances to surfaces are kept greater than, say, 0.3 to 0.5 metres. Anyway,
for other reasons, it is recommended to keep distances greater than a quarter of a wavelength, i.e.
1.3 metres at 63 Hz. A distance of 1 metre is required by ISO 3382.

161
 Measurement System
A measurement system is integrated in all editions of ODEON. It helps to obtain the impulse
response of real rooms, loudspeakers, headphones and other acoustic systems. You can either
measure or import an impulse response, and from it, the decay curves, the room acoustic parameters,
and many other relevant quantities are calculated.
The measuring system consists mainly of two buttons, available in the main toolbar:

• Measure impulse response . It makes an impulse response measurement using the

measurement system in ODEON. The measurement system uses the sine sweep method, being
very effective to suppress background noise and getting accurate results. To use this option,
you need at least a speaker and microphone.
• Load impulse response . It loads an impulse response stored on your computer. This can either
be a measurement made using ODEON or similar software measurement software, or it can be
a sketch measurement just using a hand clap or balloon with a handheld recorder like a phone.

These buttons are always active, whether or not a room has been opened. The Room Acoustic Parameter
Listbutton always remains active as well, in order to make it possible to edit parameters that
are displayed in the Load impulse response window (there are limited options in the Basics & Industrial
edition).

Note: The measuring facilities in ODEON can be used as a stand-alone tool without any associated room
model. You can record, load, edit and store impulse responses without having to perform any
simulation.

In addition to the two main buttons, several relevant tools can be found under the Tools menu from
the main menu bar:

162
What is an Impulse Response and a Sine Sweep
An impulse response is the output of a system when the input is an impulse. Mathematically, an
impulse is considered to contain all frequencies, therefore the impulse response contains the full
behaviour of the system. In room acoustics, the system is the room, the input was traditionally a
hand clap or a balloon pop, and the output is the obtained recording. However, these inputs are
only rough approximations to a true impulse, and are not as accurate compared to modern
measurement methods like the Sine Sweep (Müller, 2001). In the sine sweep, the frequencies are
played one sine tone at the time, and ODEON then translates this into an impulse response. This
is explained in more detail later. You can still use the hand claps or balloon recordings with
ODEON, but we call these “Sketch Measurements” and recommend to use the Sine Sweep method
with ODEON if accuracy is important.

Standalone measurement system

If no room is loaded, most buttons, are grayed and inactive, except the ones for the measuring
system, plus the button that leads to the video tutorials page.

In this mode you can only use ODEON as a standalone measuring system. You can either measure
a new impulse response with the Measure impulse response interface or load impulse responses
with the Load impulse response function. See further down in sections 12.3 and 12.4 for a detailed
walk-through.
If you load more than one impulse responses (SHIFT/CTRL files in the Select impulse response file(s)
dialogue), a statistics window will be displayed automatically, with various results displayed in the
same manner as in Job List>Multi-point Response.

Compare simulations and measurements

Generally, the use of the measuring system goes beyond to be just a standalone tool. One of the
strongest parts of having a measuring system inside ODEON is the possibility to compare
measurements with simulations. If a room is loaded, measurements and simulations can be

163
displayed side by side at the same time for easy comparison of results, as ODEON aims at
displaying measured and simulated results in similar graphical ways.

You can transfer measured room acoustics parameters to the Multi point response display available in
the Job List, facilitating comparison of measurements and simulations. See how in Section 12.6.

Remember: ODEON functions both as a room acoustics simulation software as well as a room acoustics
measurement software. You can compare measured with simulated results in the Multi point response
display.

Sketch Measurements
If you are not bringing the full set of measurement equipment (PC, amplifier, loudspeaker etc.),
you can record a hand clap, the popping of a balloon or a paper bag, or even the gunshot of a start
gun, e.g. using a smart phone with an application that supports recording wave files or a dedicated
portable audio recorder. The clapping hands will act as the source and the phone/recorder will act
as the receiver-microphone. The files containing impulse responses can be loaded into ODEON using
the Load impulse response button. As the energy of a handclap is limited, ODEON may not be able
to derive energy at all frequency bands (in particular lower ones) due to background noise – but
still this may be better than nothing. You can read more on how to make a sketch measurement in
the application note “Measurements with a Smart Phone” on the application notes page
https://odeon.dk/learn/application-notes/.

12.1 Technical Background

This is a chapter on the technical background of the measurements system, and for the basic user
it is not necessary to know to such detail how to do measurements.

High Quality Measurements Using the Sweep Method

The room impulse response measurements in ODEON can be performed using linear and
exponential sweep signals, which is superior to other methods in suppressing background noise,
leading to high Signal-to-Noise ratio (Müller, 2001) and high immunity to distortion. A sweep is
simply a pure sinusoidal signal, with frequency increasing monotonically in time. The sweep
method for measuring impulse responses is the least sensitive method to time variance and
distortion and it is expected to work well on most hardware systems. ODEON uses sweeps to excite
a room and record the response to this signal at a microphone (receiver) position in real time. This
sweep response is then deconvolved to give the impulse response between the source and the
microphone. This process uses upward sweeps, instead of downward, meaning that low
frequencies are played first. The energy of low frequency sound typically takes more time to decay
than the energy of high frequency sound in a room. Therefore, upward sweeps are preferable over
downward ones, in order to ensure that the decay of low frequencies is captured within the
recording of the whole sweep response, and harmonic distortion is suppressed. An extra decay
period of a few seconds is recorded after the end of the sweep, so that the remaining decay for
middle and high frequencies can be completely captured.

Octave Band Filtering

The broadband impulse response obtained by the sweep method is filtered in octave bands
between 63 Hz and 8 kHz, using 2nd order Butterworth filters, according to the IEC 61260 (1995-08)
standard. These analog Butterworth filters are implemented by digital Infinite Impulse Response
filters (IIR filters), which introduce unwanted transients in the beginning of the response.
Although the length of the impulse response of the filters is infinite, a finite effective length is used,
so 99.9% of the energy of the filtered response is included. The length of the transients in the filtered

164
response is considered equal to the effective length of the filter. Moreover, reversed filtering is
applied for decay analysis so that all the transients appear at the tail of the impulse response,
instead of the beginning, and the filter phase distortion is suppressed. ODEON automatically
excludes the transient tail when processing the impulse response.

Detection of Direct Sound

An ideal impulse response, according to geometrical acoustics (Kuttruff, 1973), would consist of an
ensemble of Dirac impulses with appropriate delays and strengths. The ideal direct sound from a
source should be a perfect Dirac impulse too, arriving at a time equivalent to the distance between
the source and the receiver. However, in reality the direct sound and all the other reflections are
not perfect Dirac impulses. In fact, each reflection consists of an onset and some decay. As the
frequency gets lower, the decay of a reflection may overlap with the onset of a subsequent
reflection. In the derivation of many of the ISO 3382 parameters it is vital to capture the amount of
early energy correctly. ODEON uses advanced algorithms for successfully detecting the energy
from the direct sound and discriminating it from the following reflections. For every impulse
response the correct onset time is indicated in the display. According to the ISO 3382 standard (ISO
3382-1,2008) the onset of the direct sound and of the whole impulse response should be counted at
least -20 dB below the peak level of the direct sound. This dB value is called Trigger level. In many
cases a lower value for the Trigger level may be desirable and this is the reason why this is an
adjustable parameter in the ODEON measuring system.

Detection of Noise Floor

Most of the impulse response recordings, whether recorded directly or obtained using the sweep
method, come with a noise tail, due to the ambient background noise and noise of the transmission
line involved (PC sound card, cables and microphone). This noise tail should be excluded before
deriving the decay curve and the ISO 3382 room acoustic parameters (ISO 3382-2, 2008). The
ODEON measurement system utilizes a modification of the algorithm by Lundeby for
automatically detecting the noise floor and truncating the impulse response at an appropriate time
(Lundeby, 1995).

12.2 Equipment to Use with the Measurement System

The ODEON measuring system is very flexible. Only elementary equipment is needed for an initial
measurement:

• Microphone (receiver).
• Loudspeaker (source).
• Sound card (audio interface).
• Amplifier(s).
• Computer.
• Cables.

These devices can be connected according to the diagram below, with several variations allowed
depending on whether an external or internal sound card is used and whether the microphone and
loudspeaker require external amplifiers or have their own type of internal amplification.

165
Microphone
A microphone is used to record the impulse response at a specific ‘receiver’ location. Many types
of microphones are suitable for our purpose, and choosing any of them is a matter of budget, ease
of use and precision. One of the most important prerequisites is that the frequency response is as
flat as possible for the octave bands between 63 Hz and 8000 Hz. If using the built-in microphone
amplifier, it has a (mixer) slider which should be set to max manually by the user.

Microphone type
According to the technology used, microphones can be divided in Condenser, Electret, Digital and
Dynamic ones.
Condenser microphones have an ultra-light diaphragm and are extremely sensitive to soft sounds,
therefore are the ideal type for room acoustic measurements in order to capture even the weakest
reflections in the room. These microphones require phantom (external) power, typically 12V or 48V,
which is often provided by external sound cards (audio interfaces).
Electret microphones require a steady electric charging (supply) to activate them. Their diaphragm
is light enough to vibrate with subtle sounds, therefore they have high sensitivity and are highly
suitable for room acoustic measurements. Usually, the electric charging comes from an internal
battery, although some models can accept phantom power from the audio interface.
Dynamic microphones are the least sensitive type. They consist of a diaphragm that can be excited
by sound waves and start vibrate. The movement generates electric signals which are then
transmitted into the rest of our equipment. Such microphones do not require electricity to operate,
but since they do not have high sensitivity, amplification is typically required to boost the recorded
impulse response in the room. Another characteristic is their typically large diaphragm which
makes the whole microphone big and not as close to an ideally ‘point’ receiver. Therefore, dynamic
microphones are not the best for room acoustic measurements.

166
Microphone class
The class defines how accurately the microphone records sounds. Class 1 is more accurate than
Class 2, because it allows less tolerance (deviation) of sound level at the different octave bands.
For example: At 1kHz, Class 1 microphones have tolerance +/- 1.1 dB, while Class 2 ones have
tolerance +/- 1.4 dB. At 20 Hz, Class 1 microphones have tolerance +/- 2.5 dB, while Class 2 ones
have tolerance +/- 3.5 dB.
Somebody can find a full comparison of the two Classes in the IEC 61672-1:2002 standard.
However, for room acoustic measurements in ODEON, the class is not crucial, simply because all
room acoustic parameters are derived for each octave band independently and without taking into
account the actual sound level that is recorded by the microphone, as long it is well above the
background noise. Even for parameters like Sound Strength G, the measuring system has to be
calibrated (see Section 12.9). Therefore, any imprecisions of the microphone will be included
automatically, as correction factors, in calibration.

Ambisonics (B-format) Microphone

Since ODEON 16, 4 input channels can be recognized when a 1 st order ambisonics (B-format)
microphone is connected. Various conventions exist for the order of the 4 ambisonics channels, but
ODEON recognizes the FuMa (Furse Malham) convention (please read more in the ODEON help,
in the Auralisation Setup). The microphone should be connected to a sound device that encodes the
4 channels directly into W, X, Y, Z, according to the FuMa convention. ODEON does not perform
the encoding. The first channel W is always the omni one.

Setting microphone amplification

It is not uncommon to have a dedicated microphone amplifier (boost) which can typically be set in
the range 0 to 30 dB, and if so, be sure to set it so a desirable response is achieved (for sufficient
SNR). If using a different setting in your measurements than those used in the calibration, be sure
to adjust the external adjustments in the measurement setup accordingly.
Often, the sound card (the audio interface) – described just below - offers sufficient microphone
amplification. Most modern sound cards are equipped with phantom power to be used with
condenser microphones. It is worth checking the capabilities of your sound card and whether they
are sufficient for your microphone requirements, before investing in an expensive dedicated
microphone amplifier.

Source (loudspeaker)
A flat frequency-response loudspeaker source can help to obtain enough Signal-to-Noise ratio
(SNR) for as many as possible octave bands.
The loudspeaker should also be as omnidirectional as possible, according to the specifications in the
ISO standard 3382-1 (2009). As reported in the standard, the source should be able to produce a
sound pressure level sufficient to provide decay curves with the required minimum dynamic
range, without contamination by background noise.

Some loudspeakers come with a built-in amplifier (active loudspeakers). They can be very useful
as no external amplification is needed.
When the source has no internal amplification (passive source) an external amplifier is needed.

Sound Card
Even built-in sound cards of a portable PCs (laptops) could be used for a simple measurement.
However, better quality of measuring equipment yields results of higher accuracy. Built-in sound
cards often use low-quality filters which introduce distortion and additional background level in
the recorded response, and in some cases there may also be electrical noise coming from other of
the PC’s components. USB audio interfaces of moderate prices, intended for gaming or music

167
recordings, can work pretty well. Power supply directly from the USB port is preferable, for
increased portability.

Analog filters of sound cards

Some cheap sound cards may have very poor realisation filter (which should cut off high
frequencies above the Nyquist frequency2). If this is the case you will notice that the sweep
reproduced over the loudspeaker does not appear to be going monotonically from low to high
frequencies. If pronounced, it is recommended to use a better interface e.g. studio-quality external
sound card. Brands like Focusrite, Motu, RME and Behringer produce external sound cards of decent
studio quality, well suited for room acoustic measurements as well, in a wide range of budgets.

Setting up your sound card correctly

Set up your sound card correctly for use with the ODEON measurement system. Many sound cards
are equipped with advanced features such as noise cancellation, echo cancellation, reverberation, which
may be useful when using internet/video conferencing or listening to music. However, such
features are highly undesirable when using the sound card for room acoustics measurements, thus
they should be disabled.
For example, many PCs come with the built-in Realtek audio device. You can access the global
settings for this device within from the Realtek Audio Manager. The noise suppression and echo
cancellation options for the recording device (microphone) should be deactivated.

SNR of sound card and microphone

The sound equipment you use for measurements (sound card, microphone and amplifier) should
provide a decent Signal-to-Noise ratio (S/N) – otherwise you will need long measurement times in
order to suppress background noise caused by the equipment. In principle, the longer the sweep
signal, the higher the suppression of the background noise. For every doubling of the sweep length
a suppression of background noise by 3dB is achieved. However, even if over-prolonging the
sweep can be an option, it is far more convenient to record at shorter times with a better SNR from
your equipment, in order to avoid accidental noise interfering during the measurement.

General recommendations
Disable system sounds in mixer
This is important so that no disturbing sounds are suddenly
played during a measurement. Since ODEON 15, the
measuring software makes use of the Windows Audio
Session Programmers Application Interface (WASAPI)
which allows better measurement quality. It bypasses the
Windows Kernel Mixer, thus avoiding noise from other
audio devices and applications on the PC, as well as the low-
quality re-sampling of the Kernel Mixer. This also allows
playback/recording in Exclusive mode by default. Therefore,
it shields the measurement from disturbing sounds by other
applications (e.g. notification from a Skype call).
If for some reason you want to deactivate Exclusive mode for
any input or output device, try to access the device
properties window and uncheck the ‘Allow application to
take exclusive control of this device’ box in the Advanced tabsheet. See the screenshot.

2
The Nyquist theorem states that sampling rate should be twice the highest frequency of concern. In room acoustic
applications the sampling frequency is 44100 Hz or 48000 Hz, which offers fine sampling up to 22050 Hz and 24000 Hz
respectively. These frequencies are well above the highest octave band of concern (8000 Hz).

168
Other methods to record impulse responses
If someone wishes to experiment with some other stimuli than the sweep signals used by ODEON
– such as gunshots or balloon/paper bag popping only a portable recorder is necessary. The
impulse response can be recorded as a .wav file and be post-processed afterwards in ODEON.

12.3 Measurement Setup Options

Measure Impulse Response Options
The Measure Impulse Response window in ODEON provides a range of options for the measurements.
Most of the options are found in the “Settings” tab in the upper right corner.

The Measure impulse response interface.

Input Levels, Test and Start:

• Input Levels: There are two level bars at the left side of the window. The blue bar shows the
current level captured by the microphone, while the red bar shows the maximum (peak) value
being captured so far. You can set the input and output levels with the two large sliders in the
center.
• Test: If you press Test, the sine sweep will loop, but not record. You can change the level of the
input and output, to find the correct level before making a recording or measurement. The rest
of the settings will be locked, until you stop the loop with the Stop button.
• Start: If you press Start, an actual measurement is made, with a sweep playback and recording.
After the sweep has been played, you are prompted to save the recorded impulse response file.
When this is done, the decay curve and acoustical parameters are calculated and displayed.

169
Input/Output level sliders
You can use the sliders for the output and input levels.
• Output: The Output will go through your soundcard to the speaker. In order to reduce the effect
of background noise, both in the room/space and from hardware, a high Output Level is
important. But be careful not to set the level of the sound source (speaker) so loud that it starts
distorting, which will affect the measurement quality. And be aware that loud noises can
damage your ears, so ear protection is often necessary if you don’t leave the room while
measuring.
• Input: The input slider controls the input level of the soundcard of the microphone. In
Windows, the input level in ODEON and the level in the Microphone properties window will be
locked together and change simultaneously. Some soundcards will have an extra slider in the
Microphone properties window called Boost. This will change the level, but not be registered by
ODEON as a change is settings, and should only be used before the first measurement. Read
more about this in the next bullet point.

Note: Changing the input level can be beneficial, especially to avoid clipping, but be aware that
setting a higher level rarely lowers the background noise. The microphone becomes more sensitive
and it picks up not only higher signal, but higher background noise too. For lowering the
background noise, use a higher output level or longer sweeps.

• Set Levels/SPL Compensation: You can also change the levels directly at your hardware or
somewhere else outside ODEON (e.g. on the soundcard or mixing software accompanying the
soundcard). However, the advantage on changing levels inside ODEON (the output and input
sliders) is that ODEON will compensate for the internal changes when evaluating the results.
For example, if you measure at the same microphone position two times with changed slider
values, the SPL levels will stay the same. In other words, if you realize that the output level
needs to be increased in the middle of a measurement session, you should use the internal
sliders. This will ensure that the SPL variation at the different microphone positions will not be
affected by this adjustment.
On the other hand, if you change the levels outside ODEON, the software will not know about
the changes, and the measurements become incomparable in terms of SPL.
If SPL, G, or STI is part of the measurements, the levels may be changed outside ODEON only
before the first measurement.
If you have changed the level outside ODEON and know by how many dB the SPL was
increased or decreased, you can compensate for this in Options > Program Setup > Measurement setup
> External adjustments > Gain.

Remember: The output device in the setup affects only the sweep signal during the measurement.
Auralisation, both in measurements and simulations is played back though the device chosen in
Windows playback devices panel.

Measurement settings
(Right part of the Measure Impulse response window)
The settings are saved when either Test or Start is pressed. If you make changes to the settings and
leave the menu again without either pressing Test or Start, the changes will be discarded.
• Escape time [ms]: this is the time from the moment you press the button until the measurements
starts. If making many loud measurements without ear protection, we recommend leaving the
room to protect your hearing. This parameter gives you time to leave the room before the
measurement starts, hence “escape time”.

170
• Silence before [ms]: added silence before the onset time of the measured impulse response.
This is applied to the resulting impulse response, not the sweep. Distortion from the speakers
can be detected in this part of the file.
• Sweep Duration [ms]: One of the most important parameters. The longer the sweep, the lower
the background noise will be. For every doubling of the sine sweep length, the background
noise will be attenuated by 3 dB. The maximum allowed duration is 80 sec.
• Impulse response length/Silence after [ms]: this parameter corresponds both to the length of
the measured impulse response and the silence recorded after the sweep. This silence is added
to ensure that the late part of the recorded sweep (the highest frequencies) has enough time to
decay in the room. We recommend using a duration that is longer than the expected
reverberation time, so that both the sound decay and the background noise are captured in the
measured impulse response.
Remember: You should always remain silent both throughout the sweep recording and the silence
after, until the measurement save dialog shows up.

Audio devices
• Output Device: Choose which output and sampling rate you are using for the measurements.
This output needs to be connected to your speakers or sound source. The sampling rate is often
controlled by the hardware used. You also have the possibility of adding an inverse filter in
order to compensate for the frequency response of the source. See in Appendix F how to obtain
an inverse filter for your device.
• Input Device: Choose the input device for the measurements. The input device needs to be
connected to (or be itself) the microphone. You can also specify an inverse filter to compensate
for the frequency response of the microphone. See in Appendix F how to obtain an inverse filter
for your device. The input can contain more than one channel. If that is the case, the following
input types are available:
o Individual channels: each channel is saved as a separate impulse response measurement.
o Omni and Figure of 8 (2 channels): channel 1 is an omnidirectional microphone and channel
2 is a Figure of 8 microphone.
o Binaural (2 channels): the two channels are measured on both ears of a dummy head.
o First Order Ambisonics (4 channels): It reads the four B-format channels W, X, Y, Z (FuMa
convention), already encoded by the recording device.
Measurements with more than one channel yield extra spatial information on the
measurements, such as the spatial parameters LF80 and LFC80 and Lj (Omni and Figure of 8,
First Order Ambisonics). The omnidirectional channel is sufficient in order to calculate
parameters such as EDT, T30, SPL, C50, C80 etc.

The two Windows icons will take you to the windows audio output and input setups.

Note: The inverse filters are saved as wave files in the application data folder, in MeasurementFilters.

Calibration and measurement conditions

• Calibration file: Here you can choose a calibration file, if you have one, to add to the
measurements directly. Calibration is needed to calculate SPL and strength parameters. All
calibration files are stored by default inside the Application data>Calibrations folder. Only for backup
purposes you can access this folder by choosing Options>Explore Application data folder.
• Measurement conditions: if the ambient conditions (temperature and relative humidity) are
known, activate the checkbox and enter the corresponding values. This will adjust the
measurement to the loaded room’s conditions (specified in Room Setup) or the default ones if no
room is loaded (specified in Program Setup).

171
 Note: you can deactivate the adjustment for measurement conditions in Program Setup by unchecking
“Normalize measurements to target conditions” in the Measurement Setup tab.

Advanced settings (optimize sweep)

• Sweep Slope: You can choose the slope type of the sine-sweep. The White (Linear) sweep has
frequency energy spread equivalent to white noise and the spectrum is a line in logarithmic axis.
The Pink (Exponential) sweep has frequency energy spread equivalent to pink noise and the
spectrum is an exponential curve. In other words, the Pink sweep provides longer playback time
for low frequencies, thus more energy at this range, while the White sweep provides longer
playback time at mid and high frequencies. For most room acoustic measurements, the Pink
(Exponential) sweep is preferred against the White (Linear) sweep. On the other hand, White (Linear)
may be preferable in measurements of sound transmission between rooms, because partitions
often provide attenuation at high frequencies. You can also make a manual slope, by choosing
Manual and specifying the Spectrum slope in the dB/Octave box.
• Frequency Range: To be more time efficient, you can choose only to measure the frequency
range of interest.
• Sweep spectrum correction: you can modify the sweep’s spectrum per frequency band by
entering corrections in dB. This will change the speed of the sweep as a function of frequency,
while the level remains constant during the measurement. This tool can be used to enhance the
measurement in octave bands with a low SNR, because of background noise or the frequency
response of the speaker used for example. Note that overall, the sweep length remains the same,
so the corrections are relative. For instance, boosting one frequency band will automatically
reduce the power in other bands. A consequence is that very large variations across bands will
only be possible for sufficiently long sweeps. If the sweep duration is too short, the variations
will be truncated.
• Automatic correction: the button Calculate from impulse response makes it possible to correct the
sweep’s spectrum automatically, using a previously measured impulse response. The correction
is based on compensating for the variations in the measured decay range across bands. Follow
the steps described below:
o Measure a first “pilot” impulse response with an uncorrected sweep. Save the resulting pilot
impulse response.
o Before carrying out the actual measurement, click on Calculate from impulse response and select
the pilot impulse response (wave file).
o Perform the actual impulse response measurement with the corrected sweep.
o Note that both the pilot and the actual measurement must be carried out in the same
conditions for optimal use.

Remember:
1. In order to calculate the spatial parameters LF(80), LFC(80) and Lj, you need an input that can
measure spatial information (either Omni+Figure of 8 or B-format probe). If you use other types, you
should uncheck the Measured option for those parameters in the Room acoustics parameters list.
2. In order to obtain meaningful values for a parameter such as SPL, it is essential that a proper Source
Power Spectrum is entered (described below) and that a system calibration has been performed. If
you do not perform any system calibration then you should uncheck the Measured option for SPL
(and other parameters that may rely on calibration) in the Room acoustics parameters list.
Press Expand tables in the upper right corner in the Room acoustic parameters list to be able to
change settings.

172
 Tip: If during the measurement, a sudden peak causes the recording to clip, ODEON terminates the
measurement showing a warning message. This way, ODEON helps you avoid storing such faulty
measurement which you may not have noticed.

Other Setup Options

In the Options>Program setup>Measurement setup important parameters can be adjusted, concerning the
measurement signal, the post processing of the impulse response, the calibration and your sound
equipment.

Parameters for detection of onset time

The Noise floor window length specifies the time interval used to detect the noise floor before the onset
time of the impulse response. When a sound file contains more than one impulse response in a row
(e.g. hand-clap recordings in a sequence), a sufficiently short Noise floor window length is required, in
order for ODEON to successfully discriminate the different impulse responses. In cases where
magnetic feedback introduces strong spikes at the beginning of the impulse response (this
measurement error happens if microphone and loudspeaker cables lie parallel and close to each
other for some distance), this setting can help the user get rid of it.
The Trigger level is the level that determines the onset time of the impulse response. Since no real
impulse response starts with a perfect DIRAC pulse, there is always some build-up of energy that
happens before the first peak of the response. ODEON first takes the square of the impulse
response so that all peaks appear positive. Then it detects the first and strongest peak of the squared
impulse response and searches backwards for a drop equal to the Trigger level. The corresponding
time at this level is the onset time of the impulse response. The default value for the Trigger level is -
40 dB. ODEON finally shifts the whole impulse response, so that the onset time is always aligned to
0 seconds.
In the following example, a squared impulse response is zoomed-in close to the first peak, which
is at around -50 dB. The onset time is shown by a vertical pink dotted line and has been located at -
90 dB, that is 40 dB below the peak, equal to the Trigger level.

Peak

Detection of Onset time when loading an impulse response. By default, the onset
time starts at 40 dB below the peak.

173
Source power spectrum
ODEON comes with two types of source power spectra that can be used with the calibration of the
measuring system: A flat frequency spectrum source, called G-ISO 3382-1 and a speech spectrum
source, called Speech-ISO 3382-3. The source type G-ISO 3382-1 should be chosen for almost all
room acoustic cases, covered in the ISO 3382-1 (auditoria, concert halls etc.). On the other hand,
source type Speech-ISO 3382-3 should be chosen if the measurement is or has been carried out in
an open plan office (ISO 3382-3 standard). Apart from the default settings, the user can define a
custom source spectrum.

Background noise
This specifies the absolute RMS dB value of the background noise in the room per octave band.
The values are obtained from the Room setup if a room has been assigned, otherwise default values
are used. The level of the background noise is used in the calculation of STI (Speech Transmission
Index) and other parameters using noise, as well as for the ISO 3382-3 results presented in the
multi-point response of ODEON.

Decay curve settings

Here you can adjust the resolution of the impulse response, as it appears at the Decay curves
display (see Load Impulse Response section in Chapter 12). When no room is assigned to ODEON
the default setting for the resolution is 3.0 ms. With a room assigned, the value is taken directly
from the Room setup, so that the measured energy response can be compared side by side to the
simulated energy response obtained in the Single Point Response.

External adjustments
The overall Gain for the sweep signal playback can be adjusted here.

12.4 Load Impulse Response

You can load an impulse response file in uncompressed .wav format by pressing the Load Impulse

Response button or by clicking Tools>Load Impulse Response (SHIFT+CTRL+L). In principle, you can load
and process any sound .wav file in the ODEON measuring system. You should be able to see even
a music signal in the ODEON measuring system editor, but calculated parameters may not make
any sense and the room acoustic parameters may not be calculated at all for such a file. The larger
the file is the longer it will take for ODEON to open. A folder called ‘Measurements’ is included in
the ODEON main folder where impulse responses are stored by default. High quality Impulse
responses can be measured using the sweep feature in ODEON if loudspeaker, amplifier,
microphone, etc. are available. It is also possible to record handclap, popping of balloon or paper
bag etc. e.g. using a smart phone or a portable audio recorder, and then load the impulse response
file into ODEON. If a file contains multiple impulse responses (such as hand claps) ODEON will
try to make use of the one providing the best Signal-to-Noise ratio (S/N). The next figure shows a
typical impulse response loaded on the Measured Impulse response window.

174
 The Measured response window in ODEON, accessed via the Load Impulse Response button.

For a multi-channel impulse response, you can select its type by using the context-sensitive menu
for any of the available Tabsheets. For example, a 4-channel measurement can either be used as a
B-format impulse response or reduced to its first channel, by choosing Raw Impulse response>Change
Impulse Response Type.

Loading a B-format impulse response

Since ODEON 16, four-channel B-format impulse responses can be recognized. The responses
should be already encoded into the W, X, Y, Z channels. Such a response will look like in the
following screenshot, with the first channel (W) being the omni, while the rest correspond to the
axes of the x, y, z coordinate system (Gerzon, 1973).

All subsequent Tabsheets show results for the W channel only, while the last Tabsheet allows
auralisation using all the channels. Since the B-format impulse response includes all necessary
directional information. The Play it! Interface allows for adjustment of the direction of the receiver.

175
12.5 Results from Measurements
Raw Impulse Response
The Raw Impulse Response displays the broadband and filtered pressure impulse response. For a
multi-channel measurement, all the channels are displayed in the graph. You can switch between
different octave-band filters by using the Up and Down arrow keys on the keyboard. A vertical dotted
purple line indicates the onset time of the impulse response. ODEON detects the strongest peak in
a recording, whether this is a typical impulse response or any sound file. If the recording consists
of a series of impulse responses (e.g. hand-claps), the onset time of the strongest impulse response
will be detected. A vertical dotted red line, unique to each octave band, indicates the truncation at
the noise floor. Whenever there is not sufficient Signal-to-Noise ratio in the impulse response
recording, the truncation time cannot be determined (safely). Instead of the truncation time, the
estimated end of the response is displayed, again by a vertical dotted red line.

You can zoom-in by clicking the left mouse button and dragging towards the lower right corner of
the graph. In that way a zoom-in rectangle is specified between certain values of the horizontal and
the vertical axes. Releasing the mouse button, the selected area is magnified to fit the whole
window. Zoom out again by clicking the left mouse button and dragging it towards upper left
corner.

Remember: Navigate through octave bands by pressing the Up and Down arrows at the keyboard or by
clicking next frequency band / previous frequency band in the Measured Response menu (that should be already
activated in the top menu bar).

Raw Decay Curve

This is a display of the squared pressure impulse response (i.e. energy impulse response) together
with the onset and truncation times. In this graph, the values of the squared pressure are in dB (re
20 μPa), which makes details more visible. However, these values will only be meaningful if the
impulse response was obtained from a calibrated simulation or measurement, otherwise the
decibels will be arbitrary. Similarly to the Raw Impulse Response, both the onset and truncation
times are displayed. In addition, a horizontal solid blue line indicates the noise floor in the impulse
response. When the noise floor of the response has a high degree of fluctuation, this solid line

176
becomes dashed, indicating that the noise floor is far from being flat, and consequently that the
estimation is not reliable.

Decay Curves
On this display the pure part of the impulse response between the onset and truncation times is
processed further. Three types of curves are displayed in the graph for each octave band:
• E, Measured: This is the energy impulse response (Raw Decay Curve) between the onset and
truncation times with a resolution specified either in the Room setup or the Measurement setup.
The display is fully compatible with the Decay Curves displayed in the simulated Point
responses in the Single Point display in ODEON Auditorium and Combined.
• E, Integrated: This is the result of the backwards Schroeder’s integration of the energy impulse
response between the onset and truncation times.
• E, Corrected: This is the E, Integrated curve with compensation for the lack of energy at the
end of the response, which is caused by the backwards Schroeder’s integration.
For a B-format impulse response, these additional curves are also displayed:
• I, Measured: This is the estimated norm of the intensity vector with the same time resolution
as the measured energy. The intensity vector is calculated using the X, Y and Z channels as an
estimate of the particle velocity vector.
• I, Integrated: This is the result of the backwards Schroeder’s integration of the intensity norm
between the onset and truncation times.

For B-format measurement, a 3D reflection view can also be opened by clicking on the button on
the right of the graph. ODEON locates intensity peaks from the measured intensity curve and
displays each peak as a spike in the 3D view. The length of the spike corresponds to the intensity
norm, its direction is the direction of the intensity vector and its colour is related to its time of
arrival. A hedgehog view displays all the intensity peaks in the time interval shown by the Decay
Curve graph. You can navigate through the intensity peaks using P and Shift+P. If a room is loaded,
it is possible to display the 3D visualization at the location of a receiver by selecting the
corresponding job. This visualization tool can be useful for detecting strong reflections and flutter
echoes.

Decay Curves (all bands)

Here the corrected energy decay curves (E, Corrected) are displayed for all bands on the same
graph. You can enable/disable the different octave bands in the right-side menu.

Energy parameters
All calculated room acoustic parameters are displayed in a table of the same format as that for the
single and multi/point responses (see Chapter 8).

Parameter curves
The room acoustic parameters are displayed in bar graphs. You can scroll among the different
parameters by using the Left and Right arrow keys in the keyboard.

Frequency Response
Displays the frequency response of the whole impulse response file, including the truncated parts.
Therefore, if you want to derive the frequency response of specific part you need to crop the
response with the Crop Impulse Response tool, described in Sec. 12.7.

Frequency Response
Displays the frequency response (spectrum) of the whole impulse response file, including the
truncated parts. Therefore, if you want to derive the frequency response of specific part you need
to crop the response with the Crop Impulse Response tool, described in Sec. 12.7.

177
Cepstrum
Displays the inverse Fourier transform of the logarithm of the Spectrum (Frequency Response).
The Cepstrum brings the signal from the frequency domain to a new domain with the unit seconds
(or preferably ms). In order to distinguish from the time domain, the independent variable in the
cepstrum is called Quefrency. It is noted that the terms Cepstrum and Quefrency have been made
by reversing the order of the first letters in spectrum and frequency.
The cepstrum analysis is very efficient to find any periodic behaviour of the spectrum. So, a single
peak in the cepstrum at quefrency Δt is an indication that the spectrum contains harmonics with
frequency spacing 1 / Δt. For more information about the cepstrum technique see (Rindel, 2016).

Spectrogram,
Displays the frequency response as a function of time. In other words, it shows the level for time
and frequency for the impulse response. It can help locating colouration issues including flutter
echoes (where some frequencies may repeat in time).

Play it! (A)(C)

Makes it possible to listen to the measured impulse response as well as to listen to a signal
convolved with the same impulse response. So, one can hear how the room measured sounds at
the given measurement position with the given source(s).
For B-format measurements, extra settings are shown at the lower half of the window. You can
choose any wave file to convolve with the synthesized impulse response. The orientation of the
listener’s head can be modified both in terms of azimuth and elevation. A binaural impulse
response is synthesized, so the you should listen to the result with headphones.

12.6 Inserting measured room acoustics parameters into

a multi-point response
If a room model has been loaded into ODEON, then it is possible to insert the room acoustics
parameters derived from a measurement file which has been opened with the Load Impulse Response.
Use the Add measured parameters (INS shortcut) or Add measured parameters and close (Ctrl+INS shortcut)
available from the Measured response menu and specify the appropriate job and receiver number in
the dialog that appears. Once the parameters have been added, you may view them in the Multi point
response (with the given job number and if the receiver number has been specified in the Source
receiver list),
available from the JobList e.g. together with simulation results (see Chapter 2.1). Even if
no simulations have been carried out, you may benefit from the Multi point response display as it can
provide results that is related to multiple receivers, e.g. statistics or results according the new ISO
3382-3 standard on open plan offices. You can toggle between simulated and measured mode in
the Multi point response by pressing the M key. The Figure below shows an example of measured and
simulated T30 in the Multi point response.

178
Simulated and measured reverberation time T30 inside the Auditorium 21 at the Technical University of Denmark. The
graph on the left shows the values at all receivers for a given octave band, while the graph on the right shows the
values at all octave bands for a given receiver. Press the Up and Down arrows to change frequency. Press R to change
receiver.

12.7 Editing Impulse Responses

Three functions are available for editing an impulse response. Cropping, Manual onset time and Manual
truncation time. All these options are available in the first two tabsheets: Raw impulse response and Raw
Decay curve.

Tip: The Raw Decay curve display is more convenient for editing an impulse response as all values have
been converted to dB, leading to a better view of the dynamic range. In addition, all reflections become
positive (same phase) and they are displayed in the upper part, making it easier to decide what is the
beginning and the end of the response.

In the next figure it is shown how to access the editing functions from the context-sensitive menu
on the main menu bar. See below for analytic descriptions of the editing functions.

Click on the Measured Response menu to see all available tools for editing and other functions.

179
Cropping
To crop an impulse response, click the left mouse button, hold down and drag a rectangle around
the part you want to crop. Release the mouse button. ODEON now zooms the cropped part to
occupy the whole window. Choose Measured Response>Crop Impulse Response (C shortcut). Now the part
will be saved as a new .wav file. Odeon asks you to choose a file name and confirm. By default, the
name of the original file is suggested, but make sure to rename it first, if you want to keep the
original file.

Manual onset time

In most cases the automatic onset time by ODEON is totally acceptable. However, since ODEON
13 it is possible to decide on your own onset time. Choose Measured Response > Manual onset time (O
shortcut). The mouse cursor becomes a cross symbol when hovering on the impulse response. Click
anywhere on the impulse response to define a new onset time for the current frequency band. If
needed, zoom the area of interest by dragging a rectangle from top left to right bottom (to zoom
out drag the opposite way). The pink dotted line defining the onset time moves now to the new
point. If Onset snap window is active the onset time tends to snap at the strongest peak. Disable Onset
snap window when complete manual control is required. When you click on another tabsheet, all
acoustic parameters are immediately updated for the corresponding band. In this way different
onset times can be defined for each band. To permanently save the changes click Measured Response
> Save Impulse response.

Manual truncation time

Similarly to the onset time you can decide where you want to truncate the impulse response by
enabling the Manual truncation time option. Choose Measured Response > Manual truncation time (T shortcut).
The mouse cursor becomes a cross symbol when hovering on the impulse response. Click anywhere
on the impulse response to define a new truncation time for the current frequency band. If needed,
zoom the area of interest by dragging a rectangle from top left to right bottom (to zoom out drag
the opposite way). The red dotted line defining the truncation time moves now to the new point.
When you click on another tabsheet, all acoustic parameters are immediately updated for the
corresponding band. In this way different onset times can be defined for each band.

Restore automatic onset/truncation times

Choose the option Restore automatic truncation times or Restore automatic onset times and truncation times to
restore the original settings by ODEON for the selected impulse response.

Save Impulse Response

When you make a manual change, the option Save impulse response becomes active so that any new
onset or truncation times can be saved with the impulse response file. Click on the option, enter a
new name, and click OK.

12.8 Making Auralisations with Impulse Responses

Auralisation features have been included in the
measuring system since ODEON 13. This allows
convolution of the measured impulse response
with an anechoic signal and listening to the
result, that is, the way the signal would sound if
it was played inside the real room. However,
quality might be limited by the microphone
used for the impulse response recording, and by
the loudspeaker used for playback.

180
Choose Play impulse response file (P) from the Measured response menu to listen to the impulse response.

Choose Auralisation (A)

to listen to an anechoic signal convolved with the impulse response. In the
Input signal section, you can make your folder and file selection, while in the Convolution section
you can choose the location of the convolved file. The file receives its name from the name of the
impulse response in use and by default it is stored in the Temp folder, meaning that it will be
deleted when shutting down Windows.

Remember: The Play impulse response file tool plays back the whole file, including the noise floor or
any other content. The Auralisation tool uses only the part between the onset and truncation times
for convolving with the signal. Therefore, manual changing of onset and truncation times will
affect auralisation.

12.9 Calibration of Measurements

So far, recording and processing of impulse responses have been described without taking
calibration into account. In order to derive level parameters such as Sound Strength (G) and Speech
Transmission Index (STI), the system set-up should be calibrated. Two calibration methods are
available in ODEON that follow the ISO 3382-1 standard (ISO 3382-2, 2008): Diffuse-field and Free-
field calibration. On the application notes page http://www.odeon.dk/application-notes you can
read more about the calibration methods described in this section. Choose Tools>Measurement
Calibration to reveal all available calibration tools.

Choose Tools>Measurement Calibration to see the available calibration tools in ODEON.

Diffuse-Field Calibration
This should be carried out in a room with long reverberation time and diffuse field, preferably a
reverberation chamber. A group of source-receiver combinations is required. According to ISO
3382-2, at least two source positions and three receiver positions (6 source-receiver combinations),
correspond to a moderate precision setting (Engineering precision). A minimum of 2 source-receiver
combinations corresponds to the lowest acceptable precision (Survey precision), while at least 12
source-receiver combinations correspond to a top precision (High precision). The impulse response
for each combination inside the reverberation chamber should be recorded and saved. Depending
on the type of Diffuse Calibration desired, One step or Two step has to be selected:

181
One step
A One step diffuse calibration corresponds to the
classical procedure described in ISO 3382-1 (ISO 3382-2,
2008). The major requirement for this type is that all
settings in the measuring equipment should stay fixed.
This means that all devices (sound card, amplifier,
microphone, cables) and gain settings (volume in the
Operating System, amplifier volume) should remain
unchanged when moving from the reverberation
chamber to the actual room under measurement.

1. Perform a number of impulse response measurements at different locations of the source and
the receiver in the reverberation chamber. Any positions are acceptable as long as the following
are taken into account:
• The height of the acoustic centre of the source should be 1.5m above the floor.
• The height of the microphone should be 1.2m above the floor.
• Different microphone positions shall be at least 2m apart.
• The distance of any microphone position to the nearest reflection surface – including the
floor – shall be at least 1m.
• No microphone position should be very close to the source.
2. Record sufficient number of impulse responses according to the accuracy desired (from at least
2, for a Survey precision, to at least 12, for a High precision.
3. Click Tools>Measurement Calibration>Diffuse-field>One step (requires fixed signal chain).
4. ODEON asks for the volume of the reverberation chamber.
5. Once this is entered you are prompted to select and load at once all impulse responses obtained
in step 1 for the same calibration (use the Ctrl-key and left mouse button).
6. ODEON will calibrate the level according to the average of the reverberation time (T30) and
Sound Pressure Level (SPL) parameters derived for each response. The calibration process
derives a correction factor for each octave band, from 63 Hz to 8000 Hz.
7. ODEON asks for a name to save the file.
8. Choose whether you want the calibration file to be active or not.

Two-step
A Two-step diffuse calibration keeps all important aspects of the classical one step diffuse
calibration, but it removes the requirement of having a fixed measuring system when moving from
the reverberation chamber to the measurement in-situ. In the paper “Applying In-situ
Recalibration for Sound Strength Measurements in Auditoria” (C. L. Christensen, 2015) the
underlying theory of this method is analyzed. All phases of the one step calibration are kept, but
an extra phase is added at the end, which ensures that the acoustic properties of the source have
been recorded. This final phase makes it easy to recover any changes in the measuring equipment
that affected the output of the source and occurred between the reverberation chamber and the
measurement in-situ. Apart from the series of impulse responses recorded normally for the diffuse
calibration method, one more impulse response measurement has to be loaded when the
microphone is placed at a known distance in front of a known direction from the source. This
happens still inside the reverberation chamber. In this way, ODEON will try to isolate the direct
sound from the source at this exact position. Later on, the measurement can be repeated in-situ. If
there is any change in the equipment that affects the power of the source, it will be reported as a
correction factor. Since the direct sound – or at least a big part of it - has to be isolated from
subsequent reflections the heights of the source and microphone should be at least 1m. Moreover,
the distance between source and receiver should be less or equal to 1 m.

Follow the procedure described in steps 1 and 2 in the One step calibration method. Then:

182
1. Click Tools>Measurements Calibrations>Diffuse-field>Two step (requires in-situ correction).
2. ODEON asks for the volume of the reverberation time.
3. Once this is entered you are prompted to select and load at once all impulse responses obtained
in step 1 for the same calibration (use the Ctrl-key and left mouse button).
4. ODEON displays all impulse responses and derives the correction values for the diffuse-field
calibration.
5. After that, the following warning message is displayed.

6. Load one or more impulse responses recorded inside the Reverberation Chamber at a well-
known position and distance in front of the source.
7. The relative levels of the source for each frequency are stored and an intermediate calibration
file is derived.
8. Save the intermediate calibration file.

In-situ correction
The intermediate calibration file can now be used to
make an in-situ correction in the room under
measurement. It cannot be used as a final
calibration file on a series of measurements. Make
one more measurement in-situ with the
microphone placed at the same position and
distance as step 8 of the Two-step calibration. Now
the final calibration file can be obtained:
1. Click Tools>Measurements Calibrations>Diffuse-field>In-situ correction.
2. ODEON asks you to load the intermediate calibration file which the complete calibration will
be applied on.
3. The same warning message as in step 5 during the Two step procedure is displayed. Heights of
source, microphone and distance between them must comply with this warning message.
4. Load one or more impulse responses recorded in-situ at a well-known position and distance in
front of the source (same as in step 6 during the Two step procedure).
5. The differences in levels for the recording in the reverberation chamber and in-situ are stored.
The final calibration file has been derived.
6. Save the final calibration file.
7. Choose whether you want the calibration file to be active or not.

Remember: After the calibration procedure is finished you are asked if you want to set the saved
calibration file as active. The active file will be displayed in the Measure impulse response (sinusoidal sweep)
window and will be applied in any subsequent measurement, unless changed in the drop-down menu.

183
Free-Field Calibration
Ideally this should be carried out in an anechoic room, but in many cases it can be carried out in a
relatively dry room. The receiver should be placed close to the source and the impulse responses
should be recorded. ODEON will try to
capture the very early part of the
response, corresponding to the direct
sound. According to the distance between
the source and the receiver and their
heights from the floor, the arrival time of
the 1st reflection is calculated so that the
impulse response is truncated at this
place. Ideally only the energy of the direct
sound should remain after the truncation.
Unfortunately, the truncation is quite
approximate at low frequencies because
the direct sound and the subsequent reflections overlap to each other, due to the long wavelengths
of the sound waves and the presence of phase shifts.

It is recommended that more than six source-receiver combinations should be used for the free-
field calibration since it is very likely that the source does not produce a perfect omni directional
pattern, but a pattern with side lobes. This means that in contrast to the diffuse field calibration,
the free-field calibration is very sensitive to receiver positions around the source. Similarly to the
diffuse calibration, you should select and load all impulse responses obtained for the same
calibration at once, so that ODEON will calibrate the level according to the average of the values.

One step
A One step free-field calibration corresponds to the classical procedure described in ISO 3382-1
(ISO 3382-2, 2008). The major requirement for this type is that any external level adjustment not
adjusted inside ODEON, is set to same level during measurement as was the case during
calibration. This includes levels set for play back level (if set to 100% it’s easy to remember and set),
recording level (if set to 100% it’s easy to remember and set) and microphone boost set inside the
Operating System as well as gain factors set on external devices used in the measurements. The
equipment itself (sound card, amplifier, microphone, cables) should also remain unchanged.

184
1. Perform a number of impulse response measurements at different locations of the source and
the receiver in the anechoic chamber. Any positions are acceptable as long as the following are
taken into account:
• The height of the acoustic centre of the source should be 1.5m above the floor.
• The height of the microphone should be 1.2m above the floor.
• Different microphone positions shall be at least 2m apart.
• The distance of any microphone position to the nearest reflection surface – including the
floor – shall be at least 1m.
• No microphone position should be very close to the source.
2. Record sufficient number of impulse responses according to the accuracy desired (from at least
2, for a Survey precision, to at least 12, for a High precision.
3. Click Tools>Measurement Calibration>Free-Field>One step (requires fixed signal chain).
4. ODEON asks for geometry values (Source Height, Receiver Height and Source-Receiver
distance).
5. Once this is entered you are prompted to select and load at once all impulse responses obtained
in step 1 for the same calibration (use the Ctrl-key and left mouse button).
6. ODEON will calibrate the level according to the files selected. The calibration process derives a
correction factor for each octave band, from 63 Hz to 8000 Hz.
7. ODEON asks for a name to save the file.
8. Choose whether you want the calibration file to be active or not.

Two-step
A Two-step diffuse calibration keeps all important aspects of the classical one step free-field
calibration, but it removes the requirement of having a fixed measuring system when moving from
the anechoic chamber to the measurement in-situ. As in the Two-step diffuse calibration method,
all phases of the one step calibration are kept, but an extra phase is added at the end, which ensures
that the acoustic properties of the source have been recorded. This final phase makes it easy to
recover any changes in the measuring equipment that affected the output of the source and
occurred between the anechoic chamber and the measurement in-situ. The tool is available at
Tools>Measurement calibration>Free-field calibration>Two-step (requires in-situ correction) for a calibration when
the main step is in an anechoic chamber.

Apart from the series of impulse responses recorded normally for the free-field calibration method,
one more impulse response measurement has to be loaded when the microphone is placed at a
known distance in front of a known direction from the source. This happens still inside the
Anechoic Chamber. In this way ODEON will try to isolate the direct sound from the source at this
exact position. Later on, the measurement can be repeated in-situ. If there is any change in the
equipment that affects the power of the source, it will be reported as a correction factor. Since the
direct sound – or at least a big part of it – has to be isolated from subsequent reflections the heights
of the source and microphone should be at least 1m. Moreover, the distance between source and
receiver should be less or equal to 1 m.

Follow the procedure described in stage 1 and 2 in the One step free-field calibration method.
1. Click Tools>Measurements Calibrations>Diffuse-field>Two step (requires in-situ correction).
2. ODEON asks for geometry values (Source Height, Receiver Height and Source-Receiver
distance).
3. Once this is entered you are prompted to select and load at once all impulse responses obtained
in step 1 for the same calibration (use the Ctrl-key and left mouse button).
4. ODEON displays all impulse responses and derives the correction values for the diffuse-field
calibration.
5. After that, the following warning message is displayed:

185
6. Load one or more impulse responses recorded inside the Anechoic Chamber at a well-known
position and distance in front of the source.
7. The relative levels of the source for each frequency are stored and an intermediate calibration file
is derived.
8. Save the intermediate calibration file.

In situ correction
The intermediate calibration file can now be used to make an in-situ correction in the room under
measurement. It cannot be used as a final calibration file on a series of measurements. Make one
more measurement in-situ with the microphone placed at the same position and distance as step 8
of the Two-step calibration. Now the final calibration file can be obtained:
1. Click Tools>Measurements Calibrations>Diffuse-field>In-situ correction.
2. ODEON asks you to load the intermediate calibration file which the complete calibration will
be applied on.
3. The same warning message as in step 7 during the Two step procedure is displayed. Heights of
source, microphone and distance between them must comply with this warning message.
4. Load one or more impulse responses recorded in-situ at a well-known position and distance in
front of the source (same as in step 8 during the Two step procedure).
5. The differences in levels for the recording in the reverberation chamber and in-situ are stored.
The final calibration file has been derived.
6. Save the final calibration file.
7. Choose whether you want the calibration file to be active or not.

Remember: After the calibration procedure is finished you are asked if you want to set the saved
calibration file as active. The file will be displayed in Options>Program setup>Measurement setup and will be
applied in any subsequent measurement when it is loaded in ODEON.

Figure 8 calibration
In ODEON 14, one more calibration method was added: the figure 8 microphone calibration. This
is necessary for measuring the Early Lateral Energy parameter, which expresses the fraction of
energy J LF arriving from lateral directions within the first 80 ms (2009):

0 , 080

p
2
L (t )dt
J LF = 0 , 005
0 , 080

p
2
(t )dt
0

In this equation:
PL (t ) is the instantaneous sound pressure of the impulse response in the room, measured with a
figure-8 pattern microphone.
P (t ) is the instantaneous sound pressure of the impulse response measured at the measurement
point with an omni-directional microphone.

186
From the foregoing equation it is shown that for such a measurement a microphone with two
channels is needed: An omni and a figure-8 channel. The sensitivities of these microphones should
be adjusted so that the peaks of the two channels coincide.

For measurement of the J LF parameter the sensitivities of an omni and a figure 8 channel should be aligned, i.e. the
maximum sensitivity value of the figure 8 channel should be equal to the sensitivity value of the omni directional
channel.

An impulse response in a fairly damped room needs to be captured with both the omni and the
figure 8 channels being active. Please make sure that 2 Channels: Omni+Figure 8 is the active input
device type in Measure Impulse Response>Type.

ODEON will try to separate the direct sound arriving at the microphone from subsequent
reflections.
1. Click Tools>Measurements Calibrations>Figure of 8 Microphone Calibration.

187
2. ODEON asks for a 2-channel impulse response.
3. Load one or more impulse responses recorded with the 2-channel microphone.
4. ODEON calculates the difference between the omni and the figure-8 channel. This difference is
then used to equalize the peaks of the corresponding sensitivities.

5. Save the calibration file.

6. Choose whether you want the calibration file to be active or not.

Assign Calibration to existing measurements

In many cases it is more convenient to
perform all impulse response recordings
first and apply calibration afterwards. This
means that impulse responses can be
obtained both in the reverberation or
anechoic chamber and in-situ at once and
later on be used for deriving the calibration
files and the final calibrated measurement
files.

Once the calibration files have been

derived with one of the methods described already, click Tools>Measurement Calibration>Assign calibration
to existing measurements. ODEON will ask for the calibration file that can be used to calibrate the
measurements. Now the in-situ recordings can be loaded. The recording steps that need to be
followed for a post-measurement calibration can be summarized as follows:

1. Perform and store as many impulse response recordings as required for a diffuse or a free-field
calibration inside an anechoic or reverberation chamber respectively.
2. If a two-step method is chosen, record an extra set of impulse responses at a specific distance
and location from the source. Write down the distance and the exact location of the microphone
relative to the source. If possible, take more than one measurement, even though the position is
the same. This will help to smooth out any random errors.

188
3. Enter the actual room under measurement (in-situ). If a two-step method is chosen, repeat the
set of measurements for the same distance – location. If any changes in the gains of the
measuring equipment have been made, they will be corrected. If nothing is changed there will
be just a null correction.
4. Record and store all impulse responses desired in the room under measurement (in-situ).

12.10 Examples of Impulse Responses – Warnings

A reliable measurement result relies on a healthy impulse response. Although it is difficult to give
a general definition on what a “healthy” impulse response is, here are some guidelines that may
help in obtaining acceptable results:

1) Sufficient signal noise ratio (at least in the octave bands of interest).

Tip: If your equipment does not allow sufficient increase of gain for obtaining a good signal to noise
ratio, set the sweep length as long as possible (eg. 60 sec). Every doubling of the sweep length results
to a suppression of the noise floor by 3 dB. As an example, increasing a sweep signal from 2 sec to 64
sec will result to a 15 dB gain in the signal to noise ratio.

2) A noise floor as flat as possible (if the noise floor is included in the impulse response) at the end
of the response is a good indication that the impulse response is not polluted by impulsive noise.
3) No presence of quantization artefacts due to poor recording resolution (input level at recording
of the impulse response should be high enough).
4) Presence of spikes/hills in the impulse response could indicate presence of impulsive noise
during the measurement no matter if the impulse response was recorded directly (using external
impulses) or obtained using the sweep method implemented in ODEON. If in doubt, it is a good
idea to make an extra measurement while still in the field to see if it is consistent. Try to make a
measurement with a much longer sweep (if using the sweep method). Normally such errors will
issue a WARNING message in the title bar of the measurement window stating the value of XI
(non-linearity parameter) is too high.

Remember: ODEON helps you making a high-quality measurement by displaying WARNING messages
in the title bar of the measurement window. Moreover, when the Signal-to-Noise ratio is not sufficient,
ODEON will display asterisks “*” instead of values for the various acoustic parameters.

In the Errors subfolder inside the Measurement folder of your ODEON installation there are examples
of bad impulse responses, due to distortion, low Signal-to-Noise ratio and electromagnetic
feedback. While Signal-to-Noise ratio or electromagnetic feedback may be something easy to track
in the measurement, distortion does not provide clear signs and might let us accept an overall pure-
quality impulse response. Examples of good and bad impulse responses are given below.

Warning messages
Depending on the quality of the measurement, ODEON may display a warning message in the title
bar of the measurement window informing of a possible concern with the measurement, prioritized
in the order listed below.

• Significant distortion products.

• Issues with the impulse/squared impulse responses (i.e. suspicious bumps).
• Problems with estimating onset and truncation.
• Unable to derive some room acoustic parameters

189
In these cases, it is recommended to perform a visual inspection of the impulse response and check
that - remember that there is more than one octave band to look at. The warning messages concern
frequencies between 125 and 4000 Hz, which are the most important according to the ISO standard
3382 (2009).

Impulse response with well-defined noise floor

The following figure displays the raw decay curve for a typical impulse response filtered at the
octave band of 1000 Hz. The particular impulse response has a well-detected noise floor (solid blue
line) with well-defined onset (pink line) and truncation times (red line). In the ODEON measuring
system, the noise floor is automatically calculated individually for each octave band. Subsequently
a truncation time is automatically derived. If another truncation time is needed, you can do so
manually following the instructions in Chapter 12.9.

A typical impulse response with a well-defined “healthy” noise floor, that is easy to be detected by ODEON.

Impulse response with fluctuating noise floor

The tail of the impulse response in the following figure is polluted by a high degree of non-flat
noise floor, full of dips and notches. This can be an indication of presence of strong background
noise during the measurement and the Signal-to-Noise ratio is insufficient for the calculation of all
room acoustic parameters. In these cases, ODEON typically displays a fluctuating noise floor with
a dashed blue line. However, in this particular case this prolonged fluctuating noise floor cannot
be detected as fluctuating and ODEON places a normal blue line instead. In such a case, it is wise
to visually inspect the fluctuating envelope of the noise floor and repeat the measurement under
less noisy conditions.

190
 C :\Odeon12C ombined\Measurements\Elmia RoundRobin2 detailed.ImpRespFile0.wav
Raw decay curve at 1000Hz

-20
-22 g
b
c
d
e
f E, Measured
-24 b
c
d
e
f
g Noise floor
-26 b
c
d
e
f
g Onset time
-28 b
c
d
e
f
g Truncation time
-30
-32
-34
-36
-38
-40
-42
-44
-46
SPL(dB)

-48
-50
-52
-54
-56
-58
-60
-62
-64
-66
-68
-70
-72
-74
-76
-78

0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1 1,1 1,2 1,3 1,4 1,5 1,6 1,7 1,8 1,9
time (seconds)
Odeon©1985-2013 Licensed to: Odeon

An impulse response with highly fluctuating noise floor.

Series of impulse responses

The next two figures show an impulse response recording containing 4 hand claps in a row. Only
one of the impulse responses has to be processed. ODEON detects the one that has the highest peak
in the broadband version – i.e. the 2nd one. The beginning of the 3rd impulse response defines the
end of the noise floor for the 2nd impulse response, so that the truncation point is correctly placed
with respect to the noise floor associated with the 2 nd impulse response. If you wish to process
another impulse response you should crop the desired part of the recording and save it as a new
file (zoom the relevant part – then save it using the c-shortcut). Once the file has been saved it is
automatically opened in the Measured Response window. You can also redefine the onset and
truncation times, as described in Chapter 12.9 and save the impulse response with a new name.

A series of hand-clap recordings in a room. The 1000 Hz octave band is shown. Hand–claps are an effective way to
obtain sketch impulse responses in rooms.

In the next figure the broadband signal is shown. Onset and truncation times are derived
automatically but the Signal-to-Noise ratio is very poor for the selected impulse response and its
length is too short, so that a stable noise floor cannot be derived. Therefore, some acoustic
parameters cannot be calculated in this case.

191
The same hand-clap recordings as in previous figure but in dB, so that the dynamic range is more visible. Broadband
signal is shown.

Impulse response with electromagnetic feedback

When long cables are used in a measurement for the microphone as well as for the loudspeaker
and the cables lie close to each other and parallel for some distance, this can lead to electromagnetic
feedback, resulting in an early peak with virtually no delay, arriving some time before the start of
the impulse response is actually detected by the microphone. This peak is undesirable and should
be removed. Changing the noise floor window length can help in placing the truncation time at a
correct place, but the onset time will be still placed wrongly at the peak of the magnetic feedback.
The best solution is to zoom the healthy part of the impulse response then save this cropped version
using the c-shortcut. Manual setting of onset time is also possible, according to Chapter 12.9. It is
of course better to avoid long parallel cables in the first place – if you are still in the field then by
all means repeat the measurement.

Electromagnetic feedback in an impulse response recording.

192
Electromagnetic feedback in an impulse response recording. Same file as in previous figure but in dB for better
visualizing the dynamic range.

Impulse response with harmonic distortion

Harmonic distortion can occur at excessive high gains in some loudspeakers. It can be a significant
problem, not easy to track and it can pollute the entire impulse response. However, part of a
distortion is visible at the beginning of the impulse response as a build-up – before arrival of direct
sound. Always inspect an impulse response for distortion and repeat the measurement if in doubt
of the quality. The following figures give examples of harmonic distortion.

The following common enhancements in sound cards can lead to distortion products. Be sure to
deactivate them:
• Echo cancellation.
• Automatic gain control.

Harmonic distortion visible as a build-up before the beginning of an impulse response. Pressure domain.

193
The impulse response shown on the previous figure but in dB. It is clearly seen that the harmonic distortion leads to a
blurred onset of the impulse response.

Impulse response with distortion at the end

Distortion at the end, which appears as a “notch” of noise may have been caused due to presence
of spikes in a measurement with the sweep method.

C :\Odeon12C ombined\Measurements\B0_04.WAV
Raw decay curve at 63Hz

-78
-80 g
b
c
d
e
f E, Measured
-82 Noise floor
b
c
d
e
f
g
-84 Onset time
b
c
d
e
f
g
-86
b
c
d
e
f
g Truncation time
-88
-90
-92
-94
-96
-98
-100
-102
-104
SPL(dB)

-106
-108
-110
-112
-114
-116
-118
-120
-122
-124
-126
-128
-130
-132
-134
-136

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4
time (seconds)
Odeon©1985-2011 Licensed to: Odeon A/S

Impulse response with a prominent notch at the end. Such a type of hill indicates that strong spikes were present during
the measurement. When using the Sweep method, spikes in the recording are converted to hills after the
deconvolution.

High Non-linearity in the decay

When the energy decay is non-linear, calculation of reverberation time may be questionable, as by
definition reverberation time is calculated for an exponential energy decay – translated to a linear
decay in a dB scale. Not only reverberation time but a group of many other parameters require an
exponential energy decay, as described in the ISO standard 3382 (2009). The XI parameter is used
to quantify the degree of non-linearity in an impulse response. Values greater than 10 ‰ indicate
a questionable non-linearity in the decay. Τhis may indicate that there is something wrong with
the captured impulse response (e.g. sudden noise during the measurement) or indeed that the

194
behaviour of the room is non-linear. As an example, assume a rectangular room with uniform low
absorption on all surfaces except from the ceiling – which is highly absorptive. The resulting decay
curve experiences a “knee” point that changes the slope and leads to a double curved decay, with
a high XI value.

In the figure below, ODEON has placed the following warning message in the title bar:
“WARNING max. XI = 158.16 at 8000 Hz” (the XI value shown is per-mille).

Impulse response with decay of high non-linearity.

Impulse response with decay of high non-linearity. Same recording as in previous figure but in dB scale, where a
normal exponential decay should look like a straight line.

Similar messages are displayed for other frequencies. If you are suspicious about the quality of
measurement and if you believe the high value of XI is due to the recording, repeat the
measurement. Check if the errors (like high XI) are reproducible. In addition, check if there is any
sign of harmonic distortion, as described previously.

195
Insufficient Signal-to-Noise ratio
In order to derive T30 at least 35 dB of valid decay must be present for each octave. Reasons for not
obtaining a valid decay curve might be:
• The captured impulse response may be too short because the impulse response length used
during the measurement was too short – in this case, make a new measurement with a longer
impulse response length.
• The Signal-to-Noise ratio was too poor, i.e. the decay disappears into the noise floor – in this
case increase the level of the loudspeaker (at the same time making sure your signal does not
clip) and/or extend the sweep length.

In the following impulse response (shown both in pressure and dB units) the Signal-to-Noise ratio
is not sufficient. According to the ISO standard 3382, the measured impulse response should be
valid at least an additional 10 dB below the lowest decay level required to derive the decay
parameter. For example, for the T30 parameter, the decay range should reach 10 dB below -35 dB,
resulting in a total decay range of 45 dB. Results may be fairly reliable as ODEON tries to
compensate for the noise floor. In the following example the decay range is less than 45 dB at 1000
Hz and therefore T30 cannot be reliably derived.

Impulse response with insufficient Signal-to-Noise ratio.

Impulse response with insufficient Signal-to-Noise ratio. Same recording as in previous figure but in dB.

196
 Tip: Every doubling of the sweep length results to a suppression of the noise floor by 3 dB. As an
example, increasing a sweep signal from 2 sec to 64 sec will result to a 15 dB gain in the signal to noise
ratio.

Impulse response between buildings

This is an example of measurement outdoors, between several buildings. In this case distant strong
spikes at high frequencies indicate the presence of prominent echoes between the buildings. At low
frequencies the effect is less prominent. In addition, the reverberation time is much longer at low
frequencies.

C :\Odeon12C ombined\Measurements\IR_BetweenBuildings2.wav
Raw decay curve at 4000Hz

50
48 g
b
c
d
e
f E, Measured
46 b
c
d
e
f
g Noise floor
44 b
c
d
e
f
g Onset time
42 b
c
d
e
f
g Truncation time
40
38
36
34
32
30
28
26
24
SPL(dB)

22
20
18
16
14
12
10
8
6
4
2
0
-2
-4
-6
-8
-10

-0.05 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95
time (seconds)
Odeon©1985-2013 Licensed to: Odeon A/S

Impulse response with prominent spikes at regular distances that indicate the presence of (flutter) echoes in the field
between the buildings. The frequency is 4000 Hz.

C :\Odeon12C ombined\Measurements\IR_BetweenBuildings2.wav
Raw decay curve at 250Hz

60
58 g
b
c
d
e
f E, Measured
56 b
c
d
e
f
g Noise floor
54 Onset time
b
c
d
e
f
g
52 Truncation time
b
c
d
e
f
g
50
48
46
44
42
40
38
36
34
SPL(dB)

32
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2

-0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2
time (seconds)
Odeon©1985-2013 Licensed to: Odeon A/S

At 250 Hz the echoes are less distinct, but the reverberation time is longer.

197
 Genetic Material Optimizer
(Combined and Auditorium editions only)

A very special feature of newer versions of ODEON, Auditorium and Combined, is a state-of-the-
art optimization tool that allows you to refine the materials in a model in order to match the real
acoustic conditions in an existing room. The tool is based on research presented in Forum
Acusticum 2014, Krakow, Poland (Christensen C.L., 2014). A genetic algorithm is utilized in the
optimization tool. The optimization process can also be called material calibration or material tuning.

Optimizing the absorption coefficients of materials inside an acoustical room model to match with
measured data is a well-known problem among room acoustic consultants. So far, the task had to
be performed manually in an iterative process of trial and error. The reason for requiring such an
optimization process is that materials in an existing room are not completely known. For example,
we might be able to know that the wall is made out of concrete but we probably cannot know the
actual type of concrete and the corresponding absorption data. Usually, 5 to 15 materials are
present in a room and most of them can be known to some degree.

Common parameters to evaluate in a calibration procedure are Early Decay Time EDT,
Reverberation Time T30, Centre Time Ts and Clarity C50, C80. Reverberation time is complementary
to Centre time and Clarity, meaning when the one increases, the other decreases and vice versa.
Among the numerous types of search algorithms, the ODEON optimization tool uses Genetic
Algorithms, which are proven to be efficient for searching in multi-dimensional search spaces
(multi-variable problems).

13.1 General Steps for Material Calibration

So far, any optimization process had to be done manually. With the Genetic Material optimizer tool in
ODEON, the largest part of the process is done automatically, but the time to set up the calculations
remains roughly the same. The general steps for a calibration are as follows:
1. Visit the room and inspect the walls. If no official data for the materials used on the natural
walls are available, an initial estimation has to be done and the closest materials in the ODEON
library should be assigned. Remember you can also add your own materials in the ODEON
library by pressing the Add Materials (CTRL+M) button in the local toolbar in the Materials list.
2. A group of impulse response measurements should be performed using the ODEON Measuring
System (see Chapter 12) or another application (see step 7). The measurements have to be done
at known positions inside the room. The ISO standard 3382 presents the guidelines regarding
the number of sources and receivers required for a given precision.
3. A virtual model of the existing room should be made in ODEON, where sources and receivers
have to be placed at the same positions as in measurements.
4. Define one multi-point response job for each source in the room in the JobList.
5. The measured impulse responses should now be loaded in ODEON using the Load impulse
response tool (see Chapter 12).
6. After inspecting the measured results and paying attention to warnings, insert the room
acoustic parameters into the multi-point responses defined in step 4. See Chapter 12 for how to
import measurements to multi-point response.

198
7. If your measured data comes from another application as a .txt file, you can import them
directly into the multi-point response.
• Open the simulated multi-point response results for a selected job.
• Click the Measured versus Simulated tabsheet.
• Choose Open/Edit measured data file.

• Now you can enter the measured data. You need to follow the format shown below:
RECEIVER 1
T(30) - 2,09 2,04 1,92 2,01 1,97 1,84 -
EDT - 2.11 2,2 2.09 2,09 2.02 1.89 -
D(50) - 0.1927 0.4857 0.4753 0.4519 0.5582 0.5768 -
C(80) - 2.36 1.14 0.76 1,10 2.59 3.15 -
Ts - 162,29 112,26 107,36 112,23 90,32 85,17 -
SPL - 10,49 10,84 10,54 9,69 8,56 8,82 -
LF(80) - 0.0062 0.0026 0.0067 0.0247 0.0458 0.0781 0.0273
STI 0.7
SPL(A) 12.2

RECEIVER 2
T(30) - 2,20 1,95 1,99 2,01 1,76 1,81 -
Example of input file. Measured results for Receiver 1 and 2 are to be imported. For Receiver 2 only the T30
parameter is available. In most cases parameters are only available for the 125 to 4000 Hz bands. For these a dash
should be assigned to the 63 and 8000 Hz bands.

Remember: Click F1 while working in the Measured versus Simulated tabsheet to get the corresponding
section in the help file, which includes instructions on how to merge measured results to simulations.

Manual Optimization
A manual optimization-calibration of a room model against a real one is based on trial and error.
Usually, two complementary acoustic parameters have to be used. Complementary parameters are
those ones which vary in different directions. A common pair is T 30 with C80. In principle as T30
increases, C80 decreases and vice versa. In addition, C80 is a rather sensitive parameter that varies
highly with position. With an initial guess of materials, a first simulation has to be carried out. If

199
measured results have been imported properly into the multi-point response, the Measured versus
Simulated tabsheet will show the degree of the agreement. If the agreement is not satisfying, some
materials need to be changed.

When small changes are needed for a material, there is no need to change or define a new material
in the Global or the Room material library. You can just modify the absorption coefficient values
on the surface list (left part), which will only affect the surfaces that use this material.

Remember: If you have to define new materials in the library you have the chance to make changes
only for the library associated with the room. Press the Toggle between local and global room library button
. The title on the top of the material library will change from Global material library to Room material
library. Now, when you add a new material or delete an existing one it will not affect the default global
library.

Change absorption coefficients here in order to leave the

material in the list unchanged.

The best approach is to focus your calibration procedure on a few most critical materials (two or
three) and try to vary them until you get the best possible match.
Example: In a classroom all materials are pretty well known except from the ceiling. An initial
estimate for the ceiling is chosen and the first simulation shows higher simulated T 30 than the
measured one. The user has to go back in the Material List and increase the absorption of the ceiling
so that the simulated T30 becomes lower and approaches the measured one better.

Genetic Algorithm Optimization

Instead of following the tedious manual optimization procedure described above, you can make
use of the ODEON’s feature called Genetic material optimizer. The tool makes use of Genetic Algorithms
and tries to modify the absorption coefficients of selected materials in the room in order to achieve
a good agreement between simulations and measurements. This optimization is done on the basis
of room acoustic parameters. The user can determine which acoustic parameters should be
involved in the process by changing their visibility in the Room Acoustic Parameter list . Before
performing any calculation, it is worth getting familiar with the basic principles of Genetic
Algorithms.

200
13.2 How Genetic Algorithms Work
Introduction
Genetic algorithms (GA) are widely used for optimization processes in diverse areas, such as
industrial design, artificial life systems and economics. GA starts with an ensemble of individuals
(chromosomes) and evolve new and improved individuals by applying principles found in
molecular genetics and biology: crossover (recombination), mutation, etc. In any stage of the
evolution the ensemble of individuals is called population and corresponds to one generation. An
individual is essentially a candidate solution to the optimization problem and normally consists of
more than one gene.

The criterion that is used from the evolution process to create an improved generation is the fitness
function. Properties of the individual that give good fitness between simulated values and target
values will have better chances of propagating into the following generations. The reason that GAs
have become popular is their ability to find useful solutions in a very complex search space having
many minima/maxima without getting stuck in the first occurring local minimum/maximum.

In our optimization problem there are eight different GAs that run independently for each octave
band. Translating the foregoing terms to our problem, an individual consists of a complete set of
absorption coefficients for a particular frequency band, corresponding to the different materials in
the room. Henceforth we shall use the term “material” for describing its absorption coefficient. One
material is one gene. The terms are shown in Table 1. All frequency-dependent GAs start with a
random pass (Monte Carlo method) where all individuals of the population are generated with
absorption coefficients which vary randomly according to a user specified search range. This can
be called 0th generation. After this pass, the evolution process is initiated by filtering out the best
individuals as parents and producing children that are likely to inherit some of the advantages
from their parents. The process continues indefinitely until the solution fully converges (criterion
is satisfied) or until the user terminates the calculation. It is unlikely that the solution will converge
fully in the acoustic optimization problem, so in practice, calculations are terminated by the user.

Target Value and Fitness Function

In order to evaluate how well a room simulation model with a given set of materials (an individual)
matches against the measured room one needs a fitness function (see Table 1). The fitness function
returns a number (fitness) to the GA that allows it to determine which individual (candidate
solution) is better than others, controlling the genetic evolution. In our problem the GA seeks for
individuals that minimize the fitness value, while in other GA problems the criterion might be a
maximization of the fitness. In principle it can be considered to evaluate directly how well the
simulated impulse responses match against the measured ones. However, there is bound to be
differences between them as neither measurements nor simulations are perfect (Christensen C.L.,
2014). Instead, we have chosen to compare how well some of the room acoustic parameters match
- as these are supposed to be good measures for important attributes of the acoustics in a room.
Indeed, a group of room acoustic parameters is one of the most important tools of the room
acoustician for the evaluation of the acoustic quality. In order to evaluate the fitness of a set of
materials, point responses for a number of source-receiver pairs are simulated and the average
deviation of a number of room acoustic parameters is calculated. The parameters are normalized
to their JND (Just Noticeable Deference) (ISO 3382-1, 2009) (e.g. 5% for reverberation time and 1 dB
for C80) so it is possible to merge different parameters into one fitness number. If the difference
between measured and simulated parameter is less than 1 JND - e.g. less than 1 dB for C80 - this is

201
fairly accurate as it is not possible for the human receiver to perceive the difference subjectively.
The fitness function used is given by the following formula:

K I

 [Par k
]
i Sim − [Parik ]Meas
 [ JND] = k =1 i =1

K I

where  is the fitness value (error between simulated and measured value) in JNDs and [Pari ]Sim
k

k
, [Pari ]Meas represent the simulated and measured acoustic parameter i for the source-receiver
combination k. K is the total number of source-receiver combinations, while I is the total number of
used acoustic parameters.
Table 1: Terms used in GA and their interpretation to the acoustic optimization problem.

GA Terms Analogue to material optimization

Gene Absorption coefficient of a material for a specific frequency band.
Chromosome Set of genes (materials) that characterize an individual.
Individual A candidate solution that consists of a list of genes (materials) - associated to a
chromosome.
Population Ensemble of Individuals - all different material combinations for one generation.
Generation A stage in the evolution process corresponding to a population.
Evolution Process of obtaining new sets of materials.
Target Measured acoustic parameters.
Fitness The error between simulated and measured data for an individual. Should be minimized.
Fitness function Calculation of fitness according the formula presented previously.

Search Space
In order to optimize the search process, it is important to limit the search space. The search space
can be limited by telling the GA that some of the absorption coefficients should only vary within
certain limits and even that some should not be changed at all.
This variation can be called search range and is given in percentage. A value of 100% would lead
to a search range from 0 to 1 absorption coefficient, regardless the initial values. A value of 0%
leads to no change at all, meaning that the material is excluded from the optimization process. A
search range between 0 and 100% gives lower and upper limits depending on the initial absorption
coefficient. Careful estimation of the search range is crucial for achieving realistic solutions. For
example, if it is suspected that two hard parallel walls may cause a flutter echo once extra
absorption is installed in the refurbished room, then it is important to restrict the absorption
coefficients and search range to low values e.g. maximum 2%. If a material is only installed on a
small surface area or it is believed to be well known it may be assigned a search range of 0% - on
the other hand there is always uncertainty when it comes to absorption coefficients, so leaving the
search range to its default of 50% may often make sense. By limiting the range of absorption
coefficients, the search process becomes more efficient as the GA will only search where there are
possible valid solutions - e.g. if it is known that there is mineral wool in the ceiling, the GA should
not specify absorption corresponding to wooden floor and vice versa. This will not only make the
search faster, but it will also prevent unrealistic solutions which match the target well but are

202
obviously wrong (wooden floor on the ceiling and mineral wool on the floor). It is recommended
that the user initially assigns materials as realistic as possible. In the ODEON Genetic material
Optimization tool it is possible to assign a search range between 0% and 100% to each material.

The interface of the Genetic material optimizer before any calculation has started. Later in this chapter an overview of
the different parameters is presented.

13.3 Perform a Material Optimization

The room Auditorium 21 at the Technical University of Denmark (DTU) will be used as an
example. A wire frame is shown in the following figure:

81 9
7 6

14
12 13
510
4 1115

P1
2

P2

Odeon©1985-2014 Licensed to: Odeon Restricted version - research and teaching only!

Load the room

Open the room Auditorium 21 at DTU.par that is pre-stored in the ODEON’s rooms folder (e.g.
C:\Odeon...\Rooms). Inspect the room in the 3DView and have a look at the Source/Receiver list . Two
sources and five receivers have been set, giving 10 combinations. Open the Material list . There are
11 different materials assigned to surfaces. The back wall, which is some kind of resonator panel
with unknown resonance frequency, and ceiling are set to 50% absorption for all frequencies. In

203
the real room it was not possible to inspect the ceiling material so an initial estimate is 2 x 13 mm
gypsum board with mineral wool back.

Have a look at the jobs defined in the Job list . The first two jobs are of particular interest, since
they have the Multi point
response enabled. The material optimization is done based on responses of
multiple receivers and comparison to measurements. Run both Job 1 and Job 2 and see the results in
the Measured versus Simulated tabsheet in the Multi point response results. The difference between
measurements and simulations for most of the parameters is significant.

Remember: In the Measured versus Simulated tab-sheet you can change receiver by hitting the R or
Shift+R key, frequencies by using the Up/Down arrows, and parameters by using the Left/Right arrows.

204
 Tip: Double - click on the graphs to get a table displaying the deviations between measurements and
simulations in JNDs (Just Noticeable Differences).

Choose Room Acoustic Parameters

Before starting the Genetic Material Optimizer, you may want to limit the optimization process to a
selected number of parameters – not the default bundle. To do so, open the Room Acoustic Parameter
list . Initially the list is collapsed showing only the active parameters (visible ones). To change
the state of each parameter, click the Expand tables button at the top right corner. The window
becomes like the one in the following figure. The visibility column appears where you can select
which parameters will be displayed in the single point response, multi-point response and grid response. The
selected parameters will be taken into consideration in the genetic material optimizer as well.
Click here to expand the list

205
 Select the active (visible) parameters

Which parameters to include

Reverberation-type parameters and fractional-type parameters are probably the best and more
stable ones to use in a material optimization process. A common (but not unique) set can be:

• EDT, T15, T20, T30 (reverberation parameters)

• Ts, C50, C80 (fractional parameters)

Reverberation-type and fractional-type parameters have a high degree of complementarity among

them. For example, if T30 increases, C80 degreases and vice versa. This makes the GA fitting process
more accurate (although more difficult), because the algorithm tries to balance counter forces.

Which parameters NOT to include

There are certain parameters in ODEON that cannot be compared with measurements, because
they are not calibrated with proper references. The optimization process can be ruined if you
include such parameters, therefore it is important to uncheck:
• All SPL parameters that have been obtained with the ODEON measuring system. At the
moment it is not possible to calibrate the system using a pistophone at a standard level (eg.
94 dB or 114 dB) at a particular frequency (eg. 250Hz or 1 kHz). Therefore, any SPL
parameter is not meaningful and typically very low compared to reality.
• G, G(early) and G(late) parameters, if the system is not calibrated for them. However, in
contrast to SPL, it is possible to calibrate ODEON for G, according to Section 12.9.
• STI, STI(male) and STI(female) parameters, if the system is not calibrated for them.
Similarly to G, there is a procedure to calibrate the system for including them. See Section
12.9.
• Although not-depended on calibration, parameters like Curvature and Dietsch Echo is better
to be avoided because they are very sensitive on measurement conditions and background
noise, therefore rather unstable.
• Clarity parameters that include background noise, such as U(50) and U(80) may be used if
background noise has been carefully set in the Room Setup of the model. Ideally it has to be
measured by an SPL meter inside the real room.

206
Open the Optimization Tool
Click the Genetic material optimizer button in the Main Toolbar or press the shortcut SHIFT+CTRL+Q. The
interface should look like in the following picture:

List of jobs List of genetic parameters Editable list of materials Original and best error displays

Live display of error Absorption coefficients

The following main areas are available in the interface:

List of Jobs: Jobs that have active sources and which has been assigned measured results are listed
here. In this example, 12 jobs have been used in the JobList, while only two of those has been assigned
measured room acoustics parameters as can be seen in the Multi point response displays of these jobs
(simulations does not need to be completed to see measured room acoustics parameters). Multi point
response’s relevant to the optimization procedure are activated by default. Keep these first two
jobs activated.

List of parameters: These parameters are for the research user of Genetic Algorithms only. As a
common user you can leave the settings at their default values. Otherwise, you can read the green
box with the title “Expert’s parameters”.

207
Editable list of materials: The list of materials as they have been assigned in the material list are
displayed here. The materials can be edited and their absorption spectrum is illustrated directly on
the right side. If absorption or measurement data are not well defined or measured for specific
bands, you can exclude them from the search process by unchecking the associated column header.
The allowed variation for each material during the optimization process is defined by the Search
Range value. When this is 0%, it means that the material will not change at all, while when it is 100%,

any absorption coefficient value can be assigned. When it is desired to exclude a material from the
search process - to reduce the search space of the algorithm and to increase the speed - all that
needs to be done is to make the Search Range value 0%. When calculations start, such a material stays
inactive.

Warning: Assigning Search Range 100% might be attractive because it allows any adjustment for a
material and can lead to a very good match between simulations and measurements. However, the
optimized material might be unrealistically different than the original material. Imagine beginning your
search with a carpet material on the floor and ending with a variation of glass.

Extra functions for materials: Two buttons allow more flexibility when editing the list of materials:
Restore coefficients. The button restores the absorption coefficients in all octave bands for the selected
material. It is useful when you have started modifying the coefficients for a material and want to
return to the original coefficients.
Copy range to other materials. If you have modified the Search Range of a material (default value is
50%), you can assign the same value to all materials by pressing this button.

Original and best error displays: The Best Fitting graph presents the original error between
simulations and measurements versus the best value so far. The error is represented in JNDs. The
Last Error Decrease shows the time elapsed since the last update of the best value so far, at the Best
Fitting graph.

Live display of error: Here the error for every individual (solution) per generation is shown.
Normally the error fluctuates highly in the beginning of an optimization process and becomes
smaller and smoother as the genetic algorithm converges.

Absorption coefficients: Three different lines of absorption coefficients are updated constantly
during a calculation.
• The Initial line corresponds to the absorption coefficients of the original material.
• The Optimized line shows the absorption coefficient that provides the best results so far.
• The Current line corresponds to the absorption coefficient of the currently running solution
(which may or may not be a good solution).
Additional to the three constantly updated curves, two fixed are displayed:
• The Low limit specifies the lowest values the corresponding material can be assigned per octave
band during the optimization process. It directly depends on the Search Range value.
• The Upper limit specifies the highest values the corresponding material can be assigned per
octave band during the optimization process. It directly depends on the Search Range value.

Tip: For a direct comparison of different materials uncheck the Auto scale option. Then 0 becomes the
minimum value and 1 becomes the maximum one.

208
Run a calculation
Now it is time to run a first calculation. Press the Start Calculation button. ODEON starts a preliminary
run, during which the original fitness values are evaluated (red bars in Best Fitting graph).
Depending on the problem this step can be extremely fast - almost unnoticeable.

Just after the preliminary run, the main Genetic Algorithm starts with the 0th generation that
involves generating completely random individuals (list of materials). However, this random
search is always restricted between the low and the upper limits specified by the search range for
each material. For example, a material that has been assigned a search range of 5% it will remain
almost the same as the original even during the random 0th generation.

After the 0th generation is finished, parents are selected with one of the methods described in the
Expert’s parameters. Then they start to evolve the next generation. From now on, the generation of
new individuals is not random. This procedure will continue until you press the Stop Calculation
button.

During the whole evolution process, all graphs in the interface are updated constantly. The
following picture shows a screenshot after about 7 generations in a calculation. You can find the
calibrated room in the Rooms folder with the name Auditorium21 at DTU_Calibrated.

Now have a look again at Multi point response for jobs No 1 and 2 in the JobList . The deviation
between measurements and simulations for most of the parameters has been reduced significantly.
Double click on the display to get a table with the differences in JNDs.

209
210
Expert’s parameters

Crossover Method: Defines the way individuals mix together to form a new individual. Two options are available:
Gene exchange and Vector mixing. With the Gene exchange option, new individuals are created completely genetically
by re-arranging.
Frequency optimization: When the individual bands option is selected, Genetic optimization is applied for each
band independently. When broadband scaling is selected, optimization is performed using an average fitness value
across all octave bands. This option aims to optimize the materials, while maintaining the exact shape of the
absorption coefficient curve.
Evolution Method: Controls how parents are selected for generating new individuals. Five options are available:
Roulette: Parents are selected based on a “roulette-wheel” selection. Each parent is assigned a portion in the
“roulette-wheel” according to its fitness. Parents with higher fitness value are more likely to be selected.
Random: Parents are selected completely randomly.
Tournament: A fixed number of individuals are selected randomly from the whole population. This is a “Tournament
Field”. Only the best parents are kept from this field. This process is repeated several times until a specified
number of the best individuals is collected.
Stochastic Tournament: The best parents from a Tournament Field are selected by selecting better fit individuals
proportional to the number of individuals per generation.
Elitist: The top n% of the population is chosen and re-chosen.
Individuals per material: A sufficient population of individuals is needed for a genetic evolution. By default, 2
individuals per material are generated. For example, if we use 10 materials in our search, 20 individuals will be
included in the procedure. Other common values are 4, 6 and so on. The values always need to be multiples of 2.
Crossover Probability: Determines which portions of two parents will be combined to create two new offspring.
Inversion Probability: Controls the likelihood that inversion will occur in a child’s chromosome, e.g., whether a
portion of the child’s chromosome will be flipped. The probability is between 0 (never) and 1 (always).
Mutation Probability: Controls the likelihood that each bit of a child’s chromosome will be mutated (flipped)
between 0 (never) and 1 (always). When 2 parents reproduce and create a new child, their chromosomes are
combined using Crossover and then the child’s chromosome may be mutated.
Elitist percent: Describes the percentage of “elite” parents used every generation for evolution. When this
percentage is 0 it means no elite parents are chosen. When this is 50%, it means that half of the parents are selected
as elite. The best 50% are chosen.

211
 Source Power Estimation
(Combined and Industrial editions only)

The Source Power Estimation tool fits Sound Pressure Level (SPL) measurement data to your ODEON
model in order to estimate the power spectrum of one or multiple active sources. Both the SPL
measurements and the sources’ powers are frequency-dependent data, presented in 8 octave bands
in ODEON.

The source power is expressed as a sound power level in dB, using the reference power of 1 pW =
10-12 W. Source power levels are typically measured in a reverberation chamber (ISO 3741, 2010),
or in a free field over a reflecting plane (ISO 3744, 2010), which is not always practical, for instance
for large equipment. It can also happen that the noise in a room originates from several sources,
which cannot be measured separately. Therefore, the Source Power Estimation tool can be a good
alternative in cases where the power of sources is a priori unknown.

14.1 Mathematical background

As a first step, the contribution of each source to each receiver is calculated by ODEON using the
initial power values input by the user. The SPL at a given receiver can be understood as a weighted
combination of these contributions.

A cost function is implemented to measure the fitness of data between the ODEON simulation and
the measurements,
𝑓(𝑥) = ‖𝐲𝑠𝑖𝑚 − 𝐲𝑚𝑒𝑎𝑠 ‖2 ,

where 𝐲𝑠𝑖𝑚 and 𝐲𝑚𝑒𝑎𝑠 represent the simulated and measured sound energies at the measurement
positions.

An independent optimisation problem is solved for each frequency band, in which the cost
function is minimized using a sequential quadratic program for non-linear constraint optimization
based on the Trust Region (TR) method. Different termination criteria are considered to stop the
search:
• Optimality condition met under certain tolerance.
• Too small step towards the minimizer or too small TR radius.
• Maximum number of iterations reached.

Note that the variables of the optimization problem are the power levels of all the active sources
in a selected job, and even though they are displayed in dB (logarithmic scale) the optimization
problem is solved in energy (linear scale).

We typically recommend more measurement points than unknown source power levels (in other
words, more receivers than sources). In that case, the problem is overdetermined, and it almost
certainly does not have an exact solution. Therefore, we do not expect a perfect fit between 𝐲𝑠𝑖𝑚
and 𝐲𝑚𝑒𝑎𝑠 . There will always be uncertainties in the measurements, especially due to background
noise, as well as in the simulations. In addition, the uniqueness of the solution is not guaranteed,
especially if too few measurements are taken. See below for further recommendations.

A sensitivity analysis is also performed in order to evaluate the stability of the solution. Each
estimated power is varied while keeping all the other powers fixed, and an upper and lower bound

212
are found within which the cost function varies less than 10 %. This means that the solution is not
much affected as long as the estimated powers remain within their respective bounds. This is
graphically displayed by error bars on the estimated sound powers.

It is expected that inaccuracies in the measured data (typically 1-2 dB) as well as imprecise
modelling of the room (i.e. unrealistic room materials, deviation in receivers location, etc.) have an
impact on the estimated powers. In this chapter, an explanation on how to interpret the results is
given, along with a best practice guide to use the tool.

14.2 Perform a source power estimation

The estimation is based on the measured SPL at different locations in the room. You will need an
accurate model of the room in ODEON to ensure meaningful results. For this guide we use the
room PowerEstimation_HallWith2Machines available in the default rooms with ODEON Industrial and
Combined installations. It is important to note that this room uses simulated SPLs as measurements.
In general, the guidelines bellow must be followed:
• The geometry of the model should agree with the actual room.
• Make sure that the surface properties are realistic. You can make use of the Genetic Material Optimizer
in order to calibrate your model (see Chapter 13).
• Use realistic models of noise sources: choose their nature accordingly (point source, surface
source, line source) and assign realistic directivities to point sources.
• For multiple active sources, the estimation works best in not too reverberant spaces, where
larger spatial variations of SPLs can be measured. In a completely reverberant space, we expect
the SPL to be approximately the same everywhere, which makes it difficult to differentiate the
different sources.

The accuracy of the results depends greatly on the choice of measurement points.
• It is advantageous to place measurement points relatively close to the sources.
• Avoid being too close to sources, as the near-field is less reliable to calculate in ODEON. For
multi-surface sources, we recommend keeping the receivers not closer than the largest
dimension of the source.
• We recommend 3-5 measurements per source.
• If possible, measure the SPL in regions which are affected mainly by one source.
• It can be beneficial to estimate the sources separately, if they can be turned off in real life.
• The receiver positions should be correctly specified in the ODEON model.

SPL measurements are typically made with a sound level meter; we assume that the measured
noises are stationary during the measurements. In your room’s ODEON model, add the SPL
measurements to your multi-point response (see Section 12.6). Below is an example of measured SPL
data formatted for ODEON:

213
 Remember: if data is missing in a frequency band, it should be indicated by a dash (-) instead of a value.

Open the Source Power Estimator and select the corresponding job in the upper drop-down list. The
items of this list show the job number, the number of active sources and the number of measured
receivers (only the jobs with measured SPL data are displayed). The 3D display shows the active
sources and the measured receivers for the selected job.

Press Calculate to estimate the source power levels. Depending on the number of sources and
receivers and the complexity of the model, the calculation may take a while.

14.3 Analysing the results

The results are presented in 4 tabs:

• All sources: shows the estimated sound power levels of all active sources for a given octave
band. You can browse through the octave bands using the up and down arrows.

214
• One source: shows the estimated sound power levels for one given source at all octave bands.
You can browse through the active sources using the left and right arrows.
• All receivers: shows the estimated and measured SPLs at all receivers for a given octave band.
You can browse through the octave bands using the up and down arrows.
• One receiver: shows the estimated and measured SPLs of one receiver at all octave bands. You
can browse through the receivers using the left and right arrows.

In the tabs showing the source powers (All sources, One source), error bars indicate the sensitivity
of the estimation. They show how much the power of the source can vary so that the resulting cost
function remains within 10 % of its optimal value. A large error bar can be explained by several
factors:
• The source does not contribute significantly to the receivers (for instance if it is behind a
partition wall). Therefore, its power does not matter and it is recommended to remove it from
the list of active sources.
• The source has a much lower power than the other ones, which should be visible in the
estimation. If this occurs in all frequency bands, it is better to leave the source out of the
estimation problem.

The two receiver tabs (All receivers, One receiver) show the fitness of data between the measured
SPLs and the simulated SPLs with the estimated source powers. Overall, you should expect a fair
agreement between them. Measurements and simulations will generally not fit perfectly, because
there can be uncertainty in the measured data (e.g., due to background noise), or differences
between the real-life scenario and the ODEON model.

215
After the calculation, you can either save the estimated powers or keep the original ones.

Note: The sound power levels in octave bands are saved in the EQ of the source, while the gain is
automatically set to 0.

14.4 Summary of the general steps

1. Identify noise sources, and determine if they can be measured one at a time. We assume noise
to be stationary.
2. Perform SPL measurements in octave bands around the sources at known positions (3-5
positions around each source).
3. Create an ODEON model of the room: pay attention to accuracy (geometry, material
properties).
4. In the Source Receiver list, add sources to model the noise sources in the room. Accuracy is also
essential (position, nature of sources, directivity for point sources).
5. In the Source Receiver list, add receivers at the same positions as in the measurements.
6. Define relevant multi-point responses in the Job List, with the proper active sources.
7. Add the measured data to the relevant multi-point response(s).
8. Open the Source Power Estimation tool .
9. Select the proper job and launch the calculation.
10. Inspect the results (estimated powers, fitness of data with measured SPLs).
11. When closing the tool, either save the new estimated powers or keep the original ones.

14.5 Best practice guide to optimise your results

Sources
• Include only the relevant sources, which contribute to noise levels at the receiver positions.
• The algorithm might struggle to distinguish two sources placed too close to each other. In that
case, consider representing the sources by only one source. Alternatively, use additional
receivers to differentiate them.
• The application is able to detect blocked sources and receivers and to exclude them from the
calculations (a warning is displayed before the calculation starts).
• If two sources have very different powers, the stronger one will mask the weaker one, so you
won’t be able to properly estimate the weaker source.

Receivers
• Ideally, each receiver should be representative of a given source. We recommend 3-5 receivers
per source, so that you have more measurement data than unknown source powers.
• A good compromise with respect to source-receiver distance must be found. Avoid the near-
field of the source, but stay relatively close to the source, so that the receiver is heavily
dependent on the source’s power.
• Receivers should not be too close to each other, otherwise the measured SPLs will be
redundant.

216
 Line Array Sources
This chapter gives an introduction to the use of the line array option in ODEON. The examples are
chosen in order to demonstrate some basic properties of line arrays, and they are not representing
recommended solutions. It is a delicate process to adjust and optimize a line arrays sound system
for a particular room, and the knowhow and technique needed for that is beyond the scope of this
manual.

15.1 Stacking the units

By stacking a number of loudspeaker units on a vertical line with a constant distance d between the
centres of the units, the first thing to note is that the splay angle changes. The sound is radiated in
a more or less concentrated beam, and the splay angle narrows in when the array gets longer. With
N units the length of the array is L = d (N-1), and this should be longer than one wavelength in
order to obtain the narrowing of the splay angle. This can be expressed by a lower limiting
frequency:
c c
fl = =
L d ( N − 1)

where c = 344 m/s is the speed of sound. However, the concentration into a single beam only works
at frequencies below the upper limiting frequency, i.e. when the distance between the units is short
compared to one wavelength:
c
fu =
d

Above fu the sound radiation breaks up into a number of directions. The near field radiation at 1
kHz is shown in Fig. 14-1, with different number of units in the array, from 1 to 11. The unit in the
example is SLS_LS8800.CF2 imported from the CLF collection of loudspeakers and the distance
between units is d = 20cm. Whereas the single loudspeaker unit spreads the sound in a wide fan,
the level decreases rapidly with distance. With increasing number of units in the array the sound
gradually concentrates in a beam with very small splay angle.

The sound radiation in the octave bands from 250 Hz to 8 kHz with 7 units is shown in Figure. 14-
2.

217
 Figure 15-1. The radiation at 1 Figure 15-2. The radiation in Figure 15-3. Similar to Figure 10-
kHz from arrays with increasing octave bands from 250 Hz to 8 2, but the array is bent, see
number of units, from a single kHz for a line array with 7 units. Table 1.
unit to 11 units (1, 3, 5, 7, 9 and
11).

In order to increase the splay angle to fit an audience area, it is usual to apply different elevation
angles to the units, and thus creating a bent array. An example is shown in Figure 14-3 and the
coordinates and elevation angles of the units are shown in Table 14-1. At high frequencies it is
obviously a problem in this example that the sound radiation splits according to the number of
units and there are gaps with poor sound radiation.

Table 15-1. Coordinates and elevation angles of the 7 units in the array example in Figure 10-3.

Transducer x y z Elevation
1 0.000 0.000 0.000 15
2 0.100 0.000 -0.200 10
3 0.150 0.000 -0.400 5
4 0.200 0.000 -0.600 0
5 0.150 0.000 -0.800 -5
6 0.100 0.000 -1.000 -10
7 0.000 0.000 -1.200 -15

15.2 Playing with delay

One advantage of the line array is the possibility to control the direction of the main lobe of sound
by means of small phase shifts, different for each of the units. So, instead of physically tilting the
loudspeaker the line array can be mounted in a vertical position and still direct the sound towards
the audience.

If the units all have the same distance d and the delay from one unit to the next is Δt, the angle of
sound radiation relative to the normal direction perpendicular to the array line is:

 c  t 
 = Arctg 
 d 

218
Figure 14-4 shows an example with Δt = 0.1 ms per unit. Figure 14-5 the same with Δt = 0.2 ms per
unit.

Note: In order to turn the beam downwards the delays should be set from 0 ms in the upper unit
to (N-1)Δt ms in the lower unit, e.g. 1.2 ms in the case of N = 7 and Δt = 0,2ms.

Figure 15-4. The radiation in octave bands from 250 Figure 15-5. Same as Figure 14-4, but with a time
Hz to 8 kHz for a line array with 7 units and a time delay 0.2 Δms per unit.
delay 0.1 Δms per unit.

At the 2 kHz band and higher frequencies (i.e. above the upper limiting frequency, 1720 Hz in this
example) a rather strong side lobe of radiation is seen at various directions slightly upwards. This
is the effect of the units being more than one wavelength apart from each other. If the phase is
shifted by one period, another angle of radiation is found. Geometrically, the angles that
correspond to i periods phase shift can be calculated from:

 c  t + i  c / f 
 i = Arctg 
 d 
where f is the frequency and i = ± 1, ± 2, etc.

The theoretical angles of radiation corresponding to this example are shown as a function of the
frequency in Table 14-2. The direction of the main lobe relative to the horizontal direction is 10° for
Δt = 0.1 ms and 19° Δt = 0.2 ms. The results for i = -1 explain the upward side lobes seen at 2 and 4
kHz in Figures 14-4 and 14-5.

219
Table 15-2. Calculated angles of radiation from the example array with d = 0.20 m. Positive angles are
downwards and negative angles are upwards. Calculated for two different values of the delay per unit, Δt.
Δt 0,1 ms Δt 0,2 ms
i 0 -1 1 i 0 -1 1
f (Hz) θ θ-1 θ1 f (Hz) θ θ-1 θ1
125 10 -86 86 125 19 -86 86
250 10 -82 82 250 19 -81 82
500 10 -73 75 500 19 -72 75
1000 10 -57 62 1000 19 -54 64
2000 10 -35 46 2000 19 -27 50
4000 10 -14 31 4000 19 -5 38
8000 10 -2 21 8000 19 7 29

Delays for individual transducers in a line array should not be confused with the overall delay
assigned to one point source or array source of many in a room in order to benefit from the Haas
effect, although the devices used in order to obtain the delays may be identical:

• Different delays assigned to individual transducers in a single line array have the purpose of
achieving the effects described in this section, with the assumption that the individual
transducers are close to each other and periodically spaced.
• For the Haas effect, on the other hand, more than one loudspeaker in the room is required, and
delays are mainly applied a whole loudspeaker (in the case of an array, the same delay is
applied to all of its transducers). The assumption is that the loudspeakers are sufficiently far
away from each other.

Example:
If transducer 1 is delayed by 0 ms and transducer 2 is delayed by 3 ms, then at t = 3 ms transducer
2 emits the same signal value as transducer 1 emitted at t = 0.

15.3 Playing with level

If the level is not the same for all the units in the array, but if it increases gradually from bottom to
top, the beam of radiation becomes asymmetric. An example is shown in Figure 14-6 with a level
increase of 2 dB per unit.

220
 Figure 15-6. The radiation in octave bands from 250 Figure 15-7. The radiation in octave bands from 250
Hz to 8 kHz for a line array with 7 units. The level Hz to 8 kHz for a line array with 7 units. The time
increases by 2 dB per unit from bottom to top. delay is 0.1 ms per unit and the level increases by 2
dB per unit from bottom to top.

15.4 Combining delay and level adjustments

With the combination of delay and level adjustments it is possible to design a sound radiation that
is asymmetric and directed off the horizontal axis, see the example in Fig. 14-7. This can be used to
create a very uniform sound level over an extended audience area.

15.5 Using the equalizer

In addition to the general equalizer for the array, there are equalizer options for each transducer.
This may be used for fine adjustments or to turn off some transducers at some frequencies. E.g. if
you want to use a shorter array at the high frequencies, you can enter a high attenuation of the
transducers in question.

15.6 Bringing the array into the room

The position of a line source in a room is similar to that for a simple point source, but with some
extra options:
• If the transducer coordinate system is Absolute it means that the x, y, z, coordinates of the array
indicate the position of the transducer with relative coordinates (0, 0, 0), if any (there doesn’t
need be a transducer in this position, but all transducer coordinates are relative to this point).
• If the transducer coordinate system is Relative hanging it means that the x, y, z, coordinates of
the array indicate the position of the transducer in the top of the array, and the coordinates of
the other transducers are relative to this.
• If the transducer coordinate system is Relative standing it means that the x, y, z, coordinates of
the array indicate the position of the transducer in the bottom of the array, and the coordinates
of the other transducers are relative to this.
• In all three cases it is possible to include an additional offset of the coordinate system. For
example, this can be used with the relative standing position to specify the distance from the
centre of the lowest transducer to the physical bottom of the array loudspeaker.

221
The direction of the array is controlled from the acoustic centre of the array, Figure 14-8. It may be
convenient to define a receiver point to be the aiming point, or the OpenGL option will let you look
into the room from the centre of the array and the aiming point is the centre of the picture (the
crossing point of the two diagonals will indicate the exact point).

0,0 0,3 0,5 0,8 1,0 metres

t1

1,0 metres t2

t3

A4 t4

0,5
t5

t6

t7
0,0

Path <m>: 0,70

Refl. order/colour:[0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10]
Time[11]
<ms>:
[>=12]
2
Odeon©1985-2009 Licensed to: Odeon A/S Dead balls: 0

Figure 15-8. An array source with 7 units (as specified in Table 14-1). The ray tracing in the room acoustic simulations
is made from the acoustical centre, shown just behind unit T4. The coordinates defining the position of the array
refers to the cross on top of the array.

222
References
Affronides S Introduction to Signal Processing [Book]. - [s.l.] : J. Prentice Hall
International, 1996.
Algazi R The CIPIC HRTF Database [Online] // The CIPIC Interface Laboratory Home
Page. - August 12, 2001. - June 16, 2011. - http://interface.cipic.ucdavis.edu/.
Allard J F and Atalla N Propagation of Sound in Porous Media: Modelling Sound
Absorbing Materials (2nd edition) [Book]. - [s.l.] : Wiley, 2009.
Anechoic Orchestral Music Recordings // Denon PG-6006. - 1988.
Bamford J S An Analysis of Ambisonics Sound Systems of First and Second Order.
MSc Thesis [Report]. - Waterloo, Ontario, Canada : University of Waterloo, 1995.
Barron M and Marshall A H Spatial Impression Due to Early Lateral Reflections in
Concert Halls: The Derivation of a Physical Measure [Article] // Journal of Sound and
Vibration, 77. - 1981. - pp. 211-232.
Barron M Auditorium Acoustics and Architectural Design [Book]. - London : E & FN
Spon, 1993.
BB93 Acoustic design of schools - performance standards [Online] // The
Department for Education (DfE), United Kingdom. - 2015. -
https://www.gov.uk/government/publications/bb93-acoustic-design-of-schools-
performance-standards.
Beranek L L and Hidaka T Sound absorption in concet halls by seats, occupied and
unoccupied, and by the hall's interior surfaces [Journal] // J. Acoust. Soc. Am. 104. -
1998. - pp. 3169-3177.
Beranek L L Concert and opera halls, how they sound [Book]. - [s.l.] : Acoustical
Society of America, 1996.
Beranek L L Concert halls and opera houses: music, acoustics and architecture
[Book]. - [s.l.] : Acoustical Society of America, 2004.
Beranek L L Music, Acoustics and Architecture [Book]. - New York : John Wiley,
1962.
Bies D A and Hansen C H Flow resistance information for acoustical design
[Journal] // Applied Acoustics. - 1980. - Vol. 13. - pp. 357-391.
Bobran H W Handbuch der Bauphysik (in German) [Book]. - Berlin : Verlag Ulstein,
1973.
Bork I A Comparison of Room Simulation Software – The 2nd Round Robin on Room
Acoustical Computer Simulation [Journal] // Acta Acustica 86. - 2000. - pp. 943-956.
Bork I Report on the 3rd Round Robin on Room Acoustical Computer Simulation –
Part II: Calculations [Journal] // Acta Acoustica United with Acoustica 91. - 2005. -
pp. 753-763.
Bradley J S and Soulodre G A Objective measures of listener envelopment
[Journal] // J. Acoust. Soc. Am. 98. - 1995. - pp. 2590-2595.
Bradley J S Predictors of speech intelligibility in rooms [Journal] // J. Acoust. Soc.
Am. 80. - 1986. - pp. 837-845.
C. L. Christensen G. Koutsouris, J. Gil, J. H. Rindel Applying In-situ Recalibration
For Sound Strength Measurements in Auditoria. [Conference] // IOA Auditorium
Acoustics. - Paris, France : [s.n.], 2015.
Christensen C L and Rindel J H Diffusion in concert halls analysed as a function of
time during the decay process [Conference] // Proceedings of the Institute of
Acoustics, Vol. 33, Pt. 2. - Dublin : [s.n.], 2011. - pp. 98-105.
Christensen C L, Nielsen G B and Rindel J H Danish Acoustical Society Round
Robin on room acoustic computer modelling [Online] // Odeon A/S. - 2008. -
http://www.odeon.dk/pdf/Classroom%20RR.pdf.
Christensen C. L [et al.] Audio effects for multi-source auralisations
[Conference] // I3DA. - Bologna, Italy : [s.n.], 2021.

223
Christensen C.L. Koutsouris G. and Holger. R Estimating absorption of materials
to match room model against existing room using a genetic algoritthm
[Conference] // Forum Acusticum 2014. - Krakow, Poland : [s.n.], 2014.
Christensen Vermund R [et al.] Fiber diameter distributions in typical MMVF wool
insulation products [Journal] // American Industrial Hygiene Association Journal. -
1993. - 5 : Vol. 54. - pp. 232-238.
Cremer L and Müller H Principles and Applications of Room Acoustics [Book]. -
London : Applied Science Publishers, 1982.
Davy John L [et al.] The average specific forced radiation wave impedance of a
finite rectangular panel [Journal] // J. Acoust. Soc. Am.. - 2014. - 2 : Vol. 136. - pp.
525-536.
Davy John L Predicting the sound insulation of single leaf walls: Extension of
Cremer's model [Journal] // J. Acoust. Soc. Am.. - 2009. - 4 : Vol. 126. - pp. 1871-
1877.
Dietsch L and Kraak W Ein objektives Kriterium zur Erfassung von Echostörungen
bei Musik- und Sprachdarbietungen (in German) [Journal] // Acustica 60. - 1986. -
pp. 205-216.
Dirac [Online] // Acoustics Engineering, Dual Input Room Acoustics Calculator. - June
16, 2011. - http://www.acoustics-engineering.com/dirac/dirac.htm.
Fasold W and Winkler H Bauphysikalische Entwurfslehre, Band 4: Bauakustik (in
German) [Book]. - Berlin : Verlag für Bauwesen, 1976.
Furse R 3D Audio Links and Information [Online]. - June 16, 2011. -
http://www.muse.demon.co.uk/3daudio.html.
Gade A C Room acoustic measurement techniques, Chapter 4 [Book Section] //
Room acoustic engineering, Note 4213. - Lyngby, Denmark : Acoustic Technology,
Technical University of Denmark, 2003.
Gade A C The influence of basic design variables on the acoustics of concert halls;
new results derived from analysing a large number of existing halls [Article] //
Proceedings of the Institute of Acoustics, Vol. 19, Pt. 3. - 1997.
Gardner B and Martin K HRTF Measurements of a KEMAR Dummy-Head
Microphone [Online]. - May 18, 1994. - June 16, 2011. -
http://sound.media.mit.edu/resources/KEMAR.html.
Gerzon M A General Metatheory of Auditory Localisation [Conference] // Preprint
3306 of the 92nd Audio Engineering Society Convention. - Vienna : [s.n.], 1992.
Gerzon Michael A. Periphony: WithHeight Sound Reproduction. [Journal]. - [s.l.] :
J. Audio Eng. Soc., 1973. - Vol. 21.
Gomez-Bolaños J, Mäkivirta A and Pulkki V Automatic Regularization Parameter
for Headphone Transfer Function Inversion [Journal] // J. Aud. Eng. Soc.. - 2016. -
10 : Vol. 64.
Harris C. M. Handbook of Acoustical Measurements and Noise Control [Book]. - New
York : McGraw-Hill, 1991.
IEC 60268-16 Sound system equipment - Part 16: Objective rating of speech
intelligibility by speech transmission index, Fourth edition [Report]. - 2011.
Ingerslev F and Petersen J Lydabsorberende materialer (in Danish) [Journal] //
Arkitektens Ugehefte no. 3. - 1953.
Ingerslev F Lærebog i bygningsakustik for Ingeniører (in Danish) [Book]. -
Copenhagen : Teknisk Forlag, 1949.
Insul [Online] // Marshall Day Acoustics. - June 16, 2011. - http://www.insul.co.nz/.
ISO 11690-1 Acoustics - Recommended practice for design of low-noise workplaces
containing machinery - Part 1: Noise control strategies [Report]. - 1996.
ISO 14257 Acoustics - Measurement and modelling of spatial sound distribution
curves in workrooms for evaluation of their acoustical performance [Report]. - 2001.
ISO 17497-1 Acoustics - Sound-scattering properties of surfaces - Part 1:
Measurement of the random-incidence scattering coefficient in a reverberation room
[Report]. - 2004.
ISO 17497-2 Acoustics - Sound-scattering properties of surfaces - Part 2:
Measurement of the directional diffusion coefficient in a free field [Report]. - 2012.

224
ISO 3382-1 // Acoustics - Measurement of room acoustic parameters - Part 1:
Performance spaces. - 2009.
ISO 3382-1 Acoustics - Measurement of room acoustic parameters - Part 1:
Performance spaces [Report]. - 2009.
ISO 3382-2 Acoustics - Measurement of room acoustic parameters - Part 2:
Reverberation time in ordinary rooms [Report]. - 2008.
ISO 3382-3 Acoustics - Measurement of room acoustic parameters - Part 3: Open
Plan Offices [Report]. - 2012.
ISO 354 Acoustics - Measurement of sound absorption in a reverberation room
[Report]. - 2003.
ISO 3741 Acoustics - Determination of sound power levels and sound energy levels
of noise sources using sound pressure - Precision methods for reverberation test
rooms [Report]. - 2010.
ISO 3744 Acoustics - Determination of sound power levels and sound energy levels
of noise sources using sound pressure - Engineering methods for an essentially free
field over a reflecting plane [Report]. - 2010.
ISO 8879 // Information processing -- Text and office systems -- Standard
Generalized Markup Language (SGML). - 1986.
ISO 9613-1 // Acoustics - Attenuation of sound during propagation outdoors, Part 1:
Calculation of the absorption of sound by the atmosphere. - 2001.
Jaouen L and Bécot F.-X. Acoustical characterization of perforated facings
[Journal] // J. Acoust. Soc. Am.. - 2011. - 3 : Vol. 129. - pp. 1400-1406.
Knudsen V O and Harris C M Acoustical Designing in Architecture [Book]. - [s.l.] :
Acoustical Society of America, 1950, 1978.
Kristensen J Sound Absorption Coefficients - Measurement, evaluation, application.
Note No. 45 (in Danish) [Report]. - Hørsholm, Denmark : Statens
Byggeforskningsinstitut, 1984.
Malham D Home page for Ambisonics and related 3-D audio research [Online] //
Music Technology Group, The University of York. - January 21, 2005. - June 16,
2011. - http://www.york.ac.uk/inst/mustech/3d_audio/.
Meyer E, Kunstmann D and Kuttruff H Über einige Messungen zur
Schallabsorption von Publikum (in German) [Journal] // Acustica 14. - 1964. - pp.
119-124.
Miki Y Acoustical properties of porous materials - Modifications of Delany-Bazley
models [Journal] // J. Acoust. Soc. Jap.. - 1990. - Vol. 11. - pp. 19-24.
Müller S., Massarani, P. Transfer-Function Measurement with Sweeps. [Journal]. -
[s.l.] : J. Audio Eng. Soc., 2001. - 6 : Vol. 49.
Music for Archimedes // CD B&O 101. - [s.l.] : B&O, Denmark, 1992.
Naylor G M and Rindel J H Odeon Room Acoustics Program, Version 2.5, User
Manual. Publication No. 49 [Report]. - Lyngby : The Acoustics Laboratory, Technical
University of Denmark, 1994.
Norcross S G., Soulodre G A. and Lavoie M C. Subjective Investigations of
Inverse Filtering [Journal] // Journal of the Audio Engineering Society. - 2004.
Ondet A M and Sueur J Development and validation of a criterion for assessing the
acoustic performance of industrial rooms [Journal] // J. Acoust. Soc. Am. 97. -
1995. - pp. 1727-31.
Oppenheim A V and Schafer R W Discrete-Time Signal Processing [Book]. -
[s.l.] : J. Prentice Hall International, 1989.
Parati L and Otondo F Comparison of Directional Sources in Simulating a Soprano
Voice [Conference] // Proceedings of the Stockholm Music Acoustics Conference
(SMAC 03). - Stockholm, Sweden : [s.n.], 2003.
Parkin P H, Humphreys H R and Cowell J R Acoustics, Noise and Buildings
[Book]. - London : Faber and Faber, 1979.
Petersen J Rumakustik (in Danish) [Book]. - Hørsholm, Denmark : Statens
Byggeforskningsinstitut, 1983.
Pierce A D Diffraction of sound around corners and over wide barriers [Journal] // J.
Acoust. Soc. Am. 55. - 1974. - pp. 941-955.

225
Press W H [et al.] Numerical Recipes in Pascal, The Art of Scientific Computing
[Book]. - [s.l.] : Cambridge University Press, 1990.
Rindel J H Acoustic Design of Reflectors in Auditoria [Conference] // Proceedings of
the Institute of Acoustics, vol. 14: Part 2. - 1992. - pp. 119-128.
Rindel J H and Christensen C L Modelling Airborne Sound Transmission between
Coupled Rooms [Conference] // Proceedings of BNAM 2008, 17-19 August. -
Reykjavik, Iceland : [s.n.], 2008.
Rindel J H Attenuation of Sound Reflections due to Diffraction [Conference] //
Proceedings of NAM-86. - Aalborg, Denmark : [s.n.], 1986. - pp. 257-260.
Rindel J H Computer Simulation Techniques for Acoustical Design of Rooms
[Article] // Acoustics Australia. - 1995. - pp. 81-86.
Rindel J H Computer simulation techniques for the acoustical design of rooms - how
to treat reflections in sound field simulation [Conference] // Proceedings of ASVA
97. - Tokyo : [s.n.], 1997. - pp. 201-208.
Rindel J H Detection of Colouration in Rooms by use of Cepstrum Technique
[Conference] // DAGA. - Aachen : [s.n.], 2016.
Rindel J H Impedance model for reflection and absorption (Manuscript) [Journal]. -
2021.
Rindel J H Modal energy analysis of nearly rectangular rooms at low frequencies
[Journal] // Acta Acustica united with Acustica. - 2015. - Vol. 101. - pp. 1211-1221.
Rindel J H Modelling the Angel-Dependent Pressure Reflection Factor [Journal] //
Applied Acoustics 38. - 1993. - pp. 223-234.
Rindel J H Sound Insulation in Buildings [Book]. - [s.l.] : CRC Press, 2018.
Rindel J H, Nielsen G B and Christensen C L Diffraction around corners and over
wide barriers in room acoustic simulations [Conference] // Proceedings of the 16th
International Congress of Sound and Vibration, 5-9 July. - Kraków, Poland : [s.n.],
2009.
Rindel J H, Nielsen G B and Christensen C L Multi source auralisation of
soundscapes in large buildings [Conference] // Proceedings of Euronoise 2009, 26-28
October. - Edinburgh, Scotland : [s.n.], 2009.
Schroeder M R Digital Simulation of Sound Transmission in Reverberant Spaces
[Journal] // J. Acoust. Soc. Am. 47. - 1970. - pp. 424-431.
Stroem S Romakustisk prosjektering (in Norwegian). Anvisning 20 [Book]. - Oslo :
Norges Byggforsknings institutt, 1979.
Vigeant M C [et al.] Multi-Channel Orchestral Anechoic Recordings for Auralizations
[Conference] // Proceedings of ISRA 2010, 29-31 August. - Melbourne, Australia :
[s.n.], 2010.
Vigran T E The acoustic properties of panels with rectangular apertures [Journal] //
J. Acoust. Soc. Am.. - 2014. - 5 : Vol. 135. - pp. 2777-2784.
Zang M, Tan K C and Er M H Three-Dimensional Sound Synthesis Based on Head-
Related Transfer Functions [Journal] // Journal of Audio Engineering Society, vol. 46
(10). - Singapore : Centre for signal processing, School of Electrical and Electronic
Engineering, Nanyang Technological University, 1998. - pp. 836-844.
Zolzer U. DAFX: Digital Audio Effects [Book]. - United Kingdom : John Wiley, 2011.

226
Appendix A: Vocabulary
The techniques of auralisation make use of many of technologies, and a lot of technical terms and
abbreviations are commonly used in the literature. Here is a short vocabulary to some of the most
used expressions - the vocabulary is not a complete description of the individual words - the
context under which the words are used are many and the subjects are rather complex.

Anechoic recording
Anechoic recordings are recordings of sound sources made without any reflections from the
surroundings contributing to the recordings. A common problem with anechoic recordings is that
they may often include too many high frequency components, because they are usually near field
recordings and because they are recorded 'on axis' where these components usually dominate.
When using such recordings with auralisation systems this may often result in unrealistic sharp 's'-
sounds especially in case of long reverberation times. Anechoic recordings are usually recorded in
an anechoic room, but semi-anechoic recordings may also be acceptable for use with auralisation
systems, this could be outdoor recordings of machinery, trains etc. or studio recordings of music.

Auralisation, auralization
The term auralisation was invented by Mendel Kleiner who gives the following definition:
Auralisation is the process of rendering audible, by physical or mathematical modelling, the sound
field of a source in a space, in such a way as to simulate the binaural listening experience at a given
position in the modelled space.

When used in ODEON, one may think of auralisation as the art of creating digital simulations of
binaural recordings in rooms (which may not even be built yet). The aim is to provide the same
three-dimensional listening experience to the listener as would be achieved in the real room at the
given receiver position with the simulated source position(s) and signals.

Binaural (recording)
Humans (usually) listen using two ears. This allows us to perceive sound as a 3D phenomenon. To
create a binaural recording, it is not enough to create a two-channel recording (stereo), since also
the effect of diffraction from the human body on sound has to be included. This is usually done by
using a dummy head with a microphone mounted at the entrance of each ear canal - this recording
may be recorded using an ordinary stereo recorder - but is now referred to as binaural. Binaural
recordings are usually played back through headphones to avoid room effects on the played sound,
as well as avoiding diffraction from the human body to be included twice (once in the recording
and additionally during playback). If one has measured or indeed simulated the BRIR's (see below)
in a room, it is possible to 'simulate' a binaural recording.

BRIR – Binaural Room Impulse Response

The BRIR is the key to binaural room acoustic auralisation. The BRIR is a set of impulse responses
detected at the left and right entrance of the ear canals of a dummy head, or indeed at blocked
entrenches of the ear canals of a (living) person residing in a room, when a sound source (or some
sound sources) has emitted an impulse. The BRIR should include all the (necessary) information

227
on receiver position and orientations, source position(s) and orientation(s), room geometry, surface
materials and the listener’s geometry (described by the HRTFs). Convolving the left channel of the
BRIR and the right channel of the BRIR with a mono signal, a binaural signal is created, which
when presented to the listener over headphones gives the impression of the three-dimensional
acoustics at a particular position in the room. It is also possible to simulate the recording of the
BRIR's, which ODEON is able to do.

DIRAC
The Dirac delta function (often called a Dirac or Dirac function) was introduced by the physicist
Paul Dirac. In room acoustics it is used for defining a reflection which has a value of 1 at the time
of the reflection (in an infinitely short interval – thus the delta). This Dirac is them multiplied by
the actual strength of the reflection e.g. in a given octave band.

Dither
Dither improves the quality of audio containing harmonic signals (not impulse responses) when
storing the signals in an integer format. Audio signals are normally saved in 16 bit integer format
which is more compact than ‘standard’ 32 bit floating point data and better supported playback
software (and it’s the format on audio CD’s). This means that software like Odeon has to quantize
the floating point signals which are results of calculations into integer values (e.g. 16.35345 becomes
16 etc.) by doing this a difference between what the signal sound have been and what it is become
occurs and this difference can be quite audible and sound bad for harmonic signals (having
sinusoidal shapes in them). Dither is a smart technique to overcome this problem. By adding a
weak noise signal to the original floating point signal before performing the quantization the
audible artefacts are reduced. As strange as it may sounds adding a noise signal reduces noise and
distortion introduced by the quantization. Still adding noise to your audio signal sounds
undesirable right! That’s were noise shaping comes into play, by applying a desired spectrum to
the dither noise it possible to put the dither noise at frequencies where we know humans are not
very sensitive to sound namely at very high and low frequencies.
If Apply dither and noise shaping is enabled in the Auralisation setup in Odeon then Odeon will apply
dithering to convolved output signals which may contain harmonics. Dithering is never applied to
impulse responses (and is not used at all if Wave result file is set to a floating point format in the
Auralisation setup. Odeon applies a dither signal with values between +/-1 with a triangular
distribution.

HRTF's – Head Related Transfer Functions

In short terms, the HRTF describes how an impulse arriving at a person /dummy head is smeared
out by diffraction phenomena from head and torso of the 'person'. While an incoming impulse is
only 1 sample long, this will result in an impulse response arriving at the right and an impulse
response arriving at the left ear, which may typically have a length (of interest) of some 2 -3
milliseconds (approximately 100 samples at a 44100 Hz sample rate, or if you prefer a length of 1
metre or so) - this is what is described by the HRTF's. A set of HRTF's used for auralisation will
typically contain a library for many different angles of incidence. The HRTF's that come with
ODEON are those made available by the CIPIC Interface Laboratory at
http://interface.cipic.ucdavis.edu/index.htm as well as those from by Bill Gardner and Keith Martin
at MIT Media Lab. at http://sound.media.mit.edu/KEMAR.html. HRTF’s will be slightly different
depending on the specific shape of a person’s body, so if possible it would be optimal to measure
and use an HRTF measured for a specific person. Then, the auralisation would turn out very
accurate for that person. Though, one should keep in mind that measuring someone’s HRTF
requires special equipment and may be a tedious and costly process. If you have the capability of

228
measuring HRTF's it is possible to import new sets for use with ODEON. To compensate for the
fact that you’ll typically not use your own HRTF’s in the auralisation process and that many
different people may indeed listen to the same auralisation, Odeon can enhance the directional
cues in the HRTF’s and does so by default. This process involves enhancing some of the spectral
differences in the HRTF while ensuring that the overall process doesn’t spectrum the auralisation
output too much. The result is typically better externalisation (out of head experience) and better
localisation.

Noise shaping
See dither.

Scattering Coefficient
It has values from 0 to 1. Defines the portion of energy that is reflected in a diffuse manner from a
surface. The rest of the energy is assumed to be reflected in a pure specular manner.

229
Appendix B: Specify Transmission
Though Walls
Since ODEON 9 transmission through walls can be simulated, taking into account multiple
transmission paths, allowing walls to have a thickness and to have different materials on either
side of the transmission wall. Any transmission properties work completely independently from
absorption. This means that you shouldn’t expect a material to absorb more sound because it has
transmission properties.
For any transmission calculation in ODEON, only 10% of the rays are let pass through a wall, while
the remaining 90% are kept inside the source room for simulation of room acoustics. Compensation
is made concerning the energy, so the calculation result in the source room is not affected by some
rays being transmitted to another room – only the absorption data for the wall is used. In general,
it is a good practice to use a higher number of rays (e.g. double the amount) when making
transmission calculations to ensure enough rays will reach the receiving room (behind the wall).

Assigning transmission data

Once a Type of a wall has been set to Transmission
in the Material List, it becomes possible to specify
transmission data, using the Edit transmission data for
surface ( button or Alt+Y shortcut). This opens a
dialog where reduction indexes can be specified
in one-third octave bands from 50 Hz to 10 kHz.
The data can be entered directly or copied from a
spreadsheet (or from a text file) using the common shortcuts Ctrl+C and Ctrl+V. For an example of
transmission, open the room Transmission rooms in the ODEON Rooms folder.

In the Specify transmission data interface you can insert transmission coefficients in dB for 1/3 octave bands.
If your data comes only in full octave bands you can insert the full octave band values and copy the data to

230
1/3 side bands by pressing the Copy center bands to 1/3 octave side bands button. The option Double sided wall
allows having the same transmission coefficient for two surfaces of different absorption.

Remember: With the transmission tool, ODEON is able to calculate the transmission of airborne sound
through a wall. It cannot calculate the structure-borne sound propagation nor flanking transmission.
ODEON requires the transmission properties (coefficients) of the wall as input data – exactly like for
absorption coefficients. When a composite structure – such as a sandwich panel - is used in ODEON
simulations, the transmission properties shall be obtained externally in four ways:
• By the datasheet from the manufacturer of the structure.
• By measurements.
• By hand calculations, using analytic formulas.
• By using software dedicated for calculating transmission coefficients of different materials (eg.
INSUL by Marshall Day Acoustics).

Copying transmission data from MsExcel

A material to be copied may contain a leading comment/name and must end by 24 floating point
values, corresponding to the 24 reduction indexes from 50 Hz to 10000 Hz. The first value is always
assumed to be for the 50 Hertz one-third octave band. So if less than 24 values are pasted, the
highest bands will be left empty, and if more than 24 values are pasted, the exceeding values are
discarded. To copy the values from MsExcel, mark the relevant cells and press Ctrl+C - to paste the
values into Transmission dialog, simply press Ctrl+V. Likewise to copy data from the Transmission
dialog to MsExcel press Ctrl+C while the dialog is the selected window in ODEON - then select the
first cell in MsExcel or whichever cell is relevant) and press Ctrl+V. Data can also be imported from
Insul Sound Insulation Prediction software (Ins11), simply by copying the spectrum for Bastian
format and pasting into ODEON.

Copying transmission data of the reduction index from MsExcel.

A wall with different material on either side and walls modeled with
thickness
Sometimes a transmission wall may be composed by two individual surfaces (two separate
surfaces in the Materials List) as shown in the illustration from the 3DBilliard display below. It is
possible for ODEON to link together such a pair of surfaces if they are (almost) parallel, allowing
transmission through walls with different absorption properties on either side. The reduction index
itself takes into consideration the wall thickness. So for a wall consisting of two parallel surfaces,
the reduction Index should only be used once. To accomplish this, it is important that:
• Transmission type is assigned to both surfaces in the Materials List.
• The same set of reduction indexes is selected on either surface in the Transmission dialog.
• Double sided wall check mark is selected for both surfaces in the Transmission dialog.
Most of the above can usually be accomplished if the Update double sided wall upon exit is checked
when Transmission data for the first wall is edited. It is recommended checking and rechecking
the data entered for transmission data before making calculations, a check may involve using the

231
3DBilliard or 3D Investigate Rays to ensure that walls do in fact transmit sound and that double-sided
walls has been set up correctly.
0 2 4 6 8 10 12 14 metres

6 metres

P3

Path <m>: 8,50

Refl. order/colour:[0] [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [>=12] Time <ms>: 25
Odeon©1985-2008 Licensed to: Odeon A/S Dead balls: 0

Figure B2: 3DBilliard display illustrating Transmission through a double sided wall – as can be seen, ODEON
understands correctly that balls should jump through the wall, from one surface to another.

The principle of calculations is shown in the figure above. Statistically 10% of the balls (rays) are
transmitted and 90% are reflected. However, this is compensated for in the calculations, so the
energy losses in the two rooms are determined by the absorption coefficients and reduction
indexes, as used.

The Transmission Rooms.par sample is installed with ODEON. This example illustrates how
transmission data is assigned to transmission walls, when transmission walls are composed from
a single surface, two surfaces and three surfaces. Once the room has been loaded into ODEON,
additional information is available in the Notes editor using the Shift+Ctrl+N shortcut. In the JobList,
some example setups have been prepared for calculations in source rooms as well as in receiver
rooms.

P1
10
9
8
20
19
18
17
16
15
14
13
12
11
25
24
23
22
4
21
P2
7

1 5

P3

2 6
27
26

P5
P4
28

Odeon©1985-2014 Licensed to: Odeon Restricted version - research and teaching only!

232
Appendix C: Importing Array
Loudspeaker Data (XML format)
The ODEON XML format for array loudspeakers allows importing array parameters, that specify
how a number of transducers are combined together to form an array loudspeaker. Arrays can be
beam-steered (a digital filter being assigned to each transducer) or a more conventional array
where each transducer is fed directly (or through an equalizer, with a delay etc.). External files
needed: a directivity pattern for each transducer. Currently the following directivity formats are
supported; Common loudspeaker format (.CF1, .CF2) and ODEON’s native format .So8.

The frequency range needed by ODEON covers the full octaves from 63 Hz to 8 kHz, however in
order to make the format as versatile as possible (e.g. for future use or for use in other programs),
frequencies from lower and higher bands can be included – ODEON will just ignore those bands.

XML (eXtensible Markup Language) has been chosen as the format for import of data for arrays
loudspeakers, because:
• It’s a standard (1986). See also http://www.w3.org/XML/
• It allows eXtending the data being exported and imported when needed, by adding new
attributes to nodes and by adding new nodes.
• Parsers for writing and reading the format are available in most (or all?) modern programming
environments.
• Even if a parser is not available to the people exporting data (e.g. Loudspeaker Manufacturer),
it is fairly easy to write XML formatted data – the import part which is more complex is left to
ODEON (or similar programs).

The layout of the XML file

To understand and learn the format, the XML files installed with ODEON e.g. in the
C:\ODEON10Combined\ArrayXML\ should be studied. This manual explains details of the
format, but most of the format is best understood from reading through the sample files. The node
(or tree) structure of XML files is complicated but best understood by studying the sample files and
looking in this manual when more details are needed. See the figures on next page to get a picture
of the tree structure.

The following XML sample files are installed with ODEON:

Monopol.xml
Dipol.xml
Quadropol.xml
Octopol.xml
Dipol_Domain_Frequency.xml

The names of the first four files are almost self-explanatory, those are the classic textbook examples
built from 1, 2, 4 and 8 sources. These sources utilize the Omni.so8 directivity pattern for all their
transducers. The four sources are all conventional arrays which have been assembled inside the
ODEON Array source editor and exported from there. The only trick is that the Invert phase option
has been applied to half of the transducers in the Dipol, Quadropol and Octopol arrays in order to
get the special behaviour of those source types. The Dipol_Domain_Frequency.xml file is essentialy
identical to the Dipol.xml and will give same results when used inside ODEON, but this file has
been defined using Domain=Frequency which is the option that can be used when defining beam-
steered arrays.

233
Figure E1: Tree structure for the Array source node - The Domain Data nodes are described separately in figure E2.

Figure E2: Tree structure with the details of the Domain Data node. This note comes in two versions depending on
weather Domain=Octave or Domain=Frequency.

234
235
Viewing the sample files in ODEON
The XML files are text files and can be viewed in a text editor such as Notepad or ODEONEdit.
Because the sample files follow the format outlined in this description, they may also be viewed in
the ODEON Array Source Editor. It is described how to view the files from within the Array Source
Editor in the “Testing if XML files…” section below.

Having loaded one of the files into the

Array Source Editor, the near field balloon,
far field balloon and a 3D_Direct display
can be studied. Note that these directivity
patterns are strongly frequency dependent
and have pronounced symmetry
properties; at high frequencies their
patterns break down and show strong
fluctuations (see figure E4), and viewing an
Octopol in the Near field balloon display
may not prove interesting because it will
never radiate any energy in its symmetry
planes, which is what is displayed in this
window (horizontal or vertical as selected).
On the other hand, the Quadropol in the
following figure gives a meaningful
display.

The Quadropol sample viewed in the near filed balloon tab in the Array source editor. To the left at 63 Hz, to the right
at 4 kHz.

Testing if XML files can be imported correctly into ODEON

The demo version of ODEON (or a full version of ODEON Auditorium or Combined) can be
used for this purpose. If you are reading this, you have probably already installed it, if this is not
the case, you may download the demo version from https://odeon.dk/downloads/free-trial/. To
model and view the array in ODEON, you only need to use a very limited number of features –
there is no need knowing all features in the ODEON software. In a full version, you may import
your array into any room. In the demo version, you should open the ArraySpeakerTestRoom.Par
room located in the “Extra Rooms>Appendix-Speaker Array” folder
(e.g. C:\ODEON11Combined\Rooms\Extra Rooms\Appendix – Speaker Array\ArraySpeakerTestRoom.Par).

The ArraySpeakerTestRoom.par room is a large box measuring 100x100x20 metres, by default all
surfaces have been assigned 100 % absorbing material as its not intended for room acoustics
predictions:
• To load this room, use the File>Open Room menu entry.
• Once loaded, open the Source receiver list (shortcut Shift+Ctrl+S)
• In the Source receiver list, define an array (shortcut A)
• In the Array source editor, import a XML array file (shortcut Ctrl+O)

236
That’s it. Hopefully the array imports without problems, otherwise load it into the ODEONEdit
editor to study the XML code. When the file is loaded, you have the chance to study the array in
various tabs of the Array source editor.

Viewing and editing XML documents

If your program writes data to be imported into ODEON then we will refer to it as WRITER,
ODEON on the other hand will be referred to as the READER (ODEON will normally be the reader,
but being able to export array data, it is a WRITER as well).

When developing export facilities for array loudspeaker data you may need to inspect or manually
edit the files. XML files can be browsed and edited in simple programs like Notepad, however, the
ODEONEdit editor will perform syntax highlighting. The ODEONEdit editor is installed with
ODEON 10 or later (including the free demo version).

Decimal point can be either “,” or “.” in the XML files. ODEON will convert the XML files upon
import to conform to the regional settings of the PC on which it’s running.
The format is case-sensitive, therefore it is important that Attribute and Node names are spelled
with upper and lower case as defined – use one of the examples installed with ODEON (and cut
and paste directly from there to your code).

Encoding and formatting of XML documents

Encoding should be set to UTF-8: <?xml version="1.0" encoding="UTF-8"?>, or in a coding
environment, XMLDocument.Encoding := 'UTF-8'. Most parsers should be able to read a number
of other encodings though. To allow reading the XML document by eye, consider to set
XMLDocument.Options := [doNodeAutoIndent], if that option is available in your programming
environment.

XML document content

Much of the data given in the XML file is optional. If not given in the XML file, ODEON will set a
default value. The nodes and Attributes of the tree-structure is described below and should be
compared with the XML-file to keep on track. See also figures E1 and E2 (page 187).

Main node (all nodes below included)

ArraySource – Main node in the XML file which holds the entire array.

Attributes to ArraySource

ODEONArrayVersion=”10.0” If this number is higher than the version number known by the READER
then the reader should not accept to read on. This version number is the version number of ODEON when
the format was last revised. That is to say, if the READER is ODEON 10.0 (or another program aware of
ODEONArrayVersion=”10.0”) then it should not accept data written in ODEONArrayVersion=”10.2”.
This on the other hand implies that if your WRITER does not use any specifications added in a later
ODEONArrayVersion then you might consider using the lowest version number possible, to make it
compatible with READERS not knowing of the higher version. READER should assume no more than 4
decimals in the version number.

Gain="8" overall in dB per octave band – defaults to 0.

Delay="0.007" Delay in seconds of the entire array.

ArrayCoordSys="Absolute" Can be "Rel. hanging", "Rel. standing" or "Absolute". Most programs
will probably export as “Absolute”. ODEON has the two other options in order to allow automatic aligning

237
and sorting of transducers relative to the upper (rel. hanging) or lower (rel. standing) transducer. The origo
of the transducers can be offset using the node CoordSysOffSet node.

FarFieldDistance="-1" In metres, -1 indicates that ODEON should make its own estimate of the far field
distance. For distances greater than FarFieldDistance ODEON will use a pre-calculated balloon (calculated
for that distance). This far field balloon is calculated and handled entirely by ODEON.

FarFieldResolution="2" Far field balloon resolution in degrees: 1, 2, 3, 5 and 10 degrees allowed.

NumberOfSubBands="6" Frequency resolution per octave band applied for phase summation in the
calculations. Can be 3,6,9… A resolution of 6 is suggested as a good compromise between calculation time
and quality of results.

Nodes to ArraySource
Position
Orientation
CoordSysOffset
EQ
Transducer (there can be multiple)

Position Node
Attributes to Position
X="1"
Y="1"
Z="1"
Uses the same coordinate system as in ODEON and as in the room to be imported into. Defaults to (1,1,1)
and can easily be changed from within ODEON.

Orientation node
Attributes to Orientation
Azimuth - +/- 180°, around vertical axis, counter-clockwise is positive.
Elevation – +/-90°, up is positive and down is negative.
Rotation"-10" – +/-180°, counter-clockwise rotation around loudspeaker axis is positive.

Nodes to Orientation
Vector

Vector node
Attributes to Vector
X="1"
Y="0"
Z="0"
Is used for defining orientation, need not be a unit vector. Above is the default orientation of an array or a
loudspeaker – right turned coordinate system assumed. Vector is just another way to specify Azimuth and
Elevation - not needed if Azimuth and Elevation have been specified – defaults to (1,0,0). READER should
allow both, WRITER should normally just use the most comfortable one.

CoordSysOffSet node
Offset of array coordinate system, see ArrayCoordSys.
Attributes to CoordSysOffSet
X="1"
Y="0"
Z="0"

EQ node
Overall equalization in dB of the array speaker, one value for each octave band.

238
Attributes to EQ
Octave_31="0"
Octave_63="0"
Octave_125="0"
Octave_250="0"
Octave_500="0"
Octave_1000="0"
Octave_2000="0"
Octave_4000="0"
Octave_8000="0"
Octave_16000="0"

ODEON will only make use of 63 to 8000 Hz octaves; other bands are ignored by ODEON. Any omitted
band defaults to 0 dB.

Transducer node (there should be at least one transducer in an array, typically there will be more)

Transducer (all nodes below included)

Attributes
Description="A descriptive text"
Gain="5"
Overall in dB per octave band – defaults to 0
Balloon="Omni.So8"
Balloon data are stored in a directory known to the reader e.g. “C:\ODEON9Combined\DirFiles”. The
balloon name may include a sub directory name (it probably should), e.g.
Balloon=“LspManufacturerName\SpeakerModel23.CF2”. Balloons can be in .CF1, .CF2 and .So8
formats.
Delay="0.012" - Delay of that transducer in seconds.

InvertPhase="false" Whether phase is inverted 180° - defaults to FALSE

SampleRate e.g. 44100 Hz or 48000 Hz. Not in use yet. Specifies the sample rate if the filter for a transducer
is an FIR filter.
PhaseCF2="false" True if directivity balloon containing phase angles should be used for the transducer. See
section on “phase balloons” at the end of this appendix.

For transducers which have been assigned a directivity pattern in the Common Loudspeaker
Format CF2 version 2, the transducer may belong to a multi part speaker, e.g. having separate
balloons for the low frequency unit and the high frequency unit in a loudspeaker. In that case, a
filter (e.g. low pass and high pass), may also be assigned to each transducer. This is what
CLFPartNo, CLFPartMax and CLFFilter is for.

CLFPartMax=”2”
How many transducers with CLF2v2 balloons belongs to this speaker (array)
CLFPartNo=”1”
Which part of a multi part CLF is this transducer (e.g. part 1 or 2 of a two-way speaker)
CLFFilter=”true”
Whether a filter assigned to the transducer in the CLF2v2 file (e.g. a low pass or a high pass filter)

Domain="Frequency"
So far options are "Frequency" for beam-steered, otherwise "Octave" (future should include "Time" in
which case an attribute, SampleRate, will be needed e.g. 44100 Hz or 48000 Hz)
Nodes
Position
Orientation
DomainData

DomainData node
There are two kinds of domain data, so far also described in figure E2. If Domain= "Octave", simple
octave band equalization is applied to each transducer. This can be used for exporting and

239
importing simple array types as they are defined inside ODEON. The format is convenient since it
is compact, and in some cases it may even be used for typing data manually into a text file.
If Domain=”Frequency”, then beaming-filters are entered as an amount of Complex numbers per
octave band. We suggest that NumberOfSubBands, an attribute of the ArraySource, is set to 6.
Attributes
DomainData has no attributes

Nodes to DomainData
EQ
or
Octave_31, Octave_63, Octave_125, Octave_250, Octave_500, Octave_1000, Octave_2000,
Octave_4000, Octave_8000, Octave_16000

EQ node
EQ node is only present if Domain="FullOctave" (Attribute of Transducer)

Attributes to EQ
Octave_31="0"
Octave_63="0"
Octave_125="0"
Octave_250="0"
Octave_500="0"
Octave_1000="0"
Octave_2000="0"
Octave_4000="0"
Octave_8000="0"
Octave_16000="0"

ODEON will only make use of 63 to 8000 Hz octaves; other bands are ignored by ODEON. Any omitted
band defaults to 0 dB.

Octave_31, Octave_63, Octave_125, Octave_250…. nodes

These nodes are only present if Domain = “Frequency”, then a number of Octave nodes are defined
– each of these Octave nodes have Sub nodes.

Nodes
Each Octave band has the centre frequency fc, defined according to ISO (31.5, 63, 125, 250, 500, 1000,
4000, 8000 and 16000 Hz). For each of the octaves there are Ns (number of Sub-bands) nodes of the
Sub type, and the center frequency of the sub bands are defined from the following algorithms:

The frequency f1 of the lowest sub-band in an octave is defined as follows:

1
fc 2 N s
f1 = 2
2

And the following frequencies fn sub-bands are defined as:

f n = f n−1 2 N s
Where n=2 to n

The 6 sub bands (Ns=6) centered around 1000 Hz have the following centre frequencies:
Sub band number – n Frequency (Hertz)
1 749.153538438340749
2 840.896415253714543

240
 3 943.874312681693497
4 1059.46309435929526
5 1189.20711500272106
6 1334.83985417003436
Frequency of nth sub band in 1000 Hz octave for Ns = 6

Results of calculations should be for full octave bands for the octave bands 63, 125, …8000 Hz
ISO suggests that center sub octave frequencies should be derived from 1000 Hz. For an uneven
number of bands e.g. 1/3rd octave bands this gives us one band below, one band at, and one band
above the center frequency of the Octave center frequency – so the 3 sub-octaves are in balance
relative to the octave band – however for an even number of sub-band frequencies thing becomes
unbalanced if using ISO frequencies; there will be 2 sub-bands below and 3 sub-bands above the
full octave band centre frequency (or opposite 3 below, 2 above)

Attributes
Re
Img
The energy of the transducer is emitted in Ns Sub nodes. The center frequencies of the Sub bands
are centered around the full octave band centre frequency, in order best to simulate the energy in
each octave band. If Ns is odd, e.g. 5 then these frequencies will coincide with ISO centre
frequencies. Be aware though that the centre frequencies used will differ if an even number (e.g. 6)
is used.

Getting the voltage and energy right

Values are in tension [V]. The values should be reduced by a factor which compensates for the
NumberOfSubBands. Ns used if Ns sub-bands are used, then the Re and Img attributes of each node
1
should be multiplied by a factor .
√Ns
If there were 6 sub-nodes in each band, and Re= "1.15534266201273234" and Img= "0.0" for each of
them, then the equivalent voltage for one sub-band representing the total power in that band would
be 1.15534266201273234xSqrt(6)=2.83 V. 2.83 V is the standard voltage for which sensitivity at 1
metre is given for loudspeakers, so with this setting it will reproduce the sensitivity values at a
distance of 1 metre, see the Dipol_Domain_Frequency.xml sample. 2.83 V may seem like an
arbitrary value, but it is generally agreed upon as ‘standard voltage´– this is the voltage at which
an 8Ω speaker deliver 1W. The reason why the industry has moved from 1W to 2.83V?
Loudspeakers rarely have a frequency-linear impedance (of 8Ω), whereas an amplifier is actually
fairly capable of supplying constant voltage.

Position node
Attributes
X="1"
Y="2"
Z="-0,324"
Position of Source of transducer origin given in metres. Position defaults to (1,1,1)

Orientation node
Attributes
Azimuth="27" - +/- 180°, around vertical axis, counter-clockwise is positive
Elevation="-10" – +/-90°, up is positive and down is negative
Rotation"-10" – +/-180°, counter-clockwise rotation around loudspeaker axis is positive
Nodes
Vector

Vector node
Attributes

241
X="0,993768018134063"
Y="0,0694910287794796"
Z="0,0871557402186753"

242
Appendix D: Inverse filter design
The inverse filter tool makes it possible to create a regularised inverse filter for a given device’s
impulse response.

Principle of inverse filtering. Without filtering, the signal can be coloured when passing through a device (e.g.,
headphones, a microphone or a loudspeaker). The role of the inverse filter is to compensate for the device so that it
becomes “transparent” in the chain.

The tool is used for two different applications in ODEON:

• Auralisation: the inverse filter compensates for the headphones’ response, so that auralisations
sound as without headphones. The tool is accessed from the Auralisation setup and requires a
measured impulse response for a selected pair of headphones (1 or 2 channels).
• Measurements: the inverse filter compensates for the source (loudspeaker) or the receiver
(microphone) during a room impulse response measurement. For that purpose, go to Tools >
Create Inverse filter (.wav).

The inverse filter can be created for one or two channels. The menu enables you to visualise, control
and modify the inverse filter, both in the time and the frequency domain. Nevertheless, the process
is kept as automatic as possible.

Calculation principles
For the inverse filter to be effective, the convolution between the headphone impulse response and
the inverse filter should be close to a Dirac function. In the frequency domain, this is equivalent to
the product of the two transfer functions being equal to 1.

The inverse filter is calculated in the frequency domain, which is obtained by applying a Fast
Fourier Transform (FFT) to the device’s measured impulse response. The idea is then to invert the
device’s transfer function in the frequency domain. However, a simple inversion is not
recommended because it may be erroneous in frequency regions where the response level is low,
as it may be contaminated by background noise. In particular, if the frequency response has clear
notches, the inverse filter might contain peaks that will translate into ringing artefacts.

A way of circumventing that issue is to make use of regularisation. We follow a methodology from
(Gomez-Bolaños, et al., 2016), which proposes an automatic regularisation, both on notches and on
level:

243
• Notches are identified by comparing the device’s frequency response to a smoothed version of
it.
• In terms of level, we regularise the response when its level is below its maximum amplitude
minus 20 dB.

The equation for the inverse filter 𝐺(𝜔) is

𝐻 ∗ (𝜔)
𝐺(𝜔) = .
|𝐻(𝜔)|2 + 𝛼 2(𝜔) + 𝜎 2 (𝜔)

• 𝐻(𝜔) is the device’s frequency response. 𝐻 ∗ (𝜔) is its complex conjugate.

• 𝛼(𝜔) is a regularisation term related to the level of 𝐻(𝜔).
• 𝜎(𝜔) is a regularisation term related to the notches of 𝐻(𝜔).

In regions where |𝐻(𝜔)| is small, 𝛼 and/or 𝜎 dominate the denominator and prevent 𝐺 from
̃ over half
reaching too high values. Both 𝛼 and 𝜎 make use of a smoothed frequency response 𝐻
octaves, which basically represents |𝐻| without notches.

The resulting multiplication between the device’s frequency response and the inverse filter’s is
almost flat, except notches, which are not perceived by the human ear if they are narrow enough.

This is an automatic process, so you don’t need to specify any regularisation parameters.

Notch regularisation: calculation of 𝝈(𝝎)

̃(𝜔).
The term 𝜎(𝜔) compares the device’s frequency response 𝐻(𝜔) with its smoothed magnitude 𝐻
It is calculated as follows:

̃(𝜔) − |𝐻(𝜔)| 𝑖𝑓 𝐻
̃(𝜔) ≥ |𝐻(𝜔)|
𝜎(𝜔) = {𝐻
0 𝑒𝑙𝑠𝑒.

When the smoothed response is larger than the original response (indicative of a notch), we use
the difference between the two responses as a value of 𝜎(𝜔). The larger the notch, the larger the
value. In other cases, 𝜎 is set to 0 (i.e., no regularisation).

Level regularisation: calculation of 𝜶(𝝎)

A threshold is defined corresponding to the maximum of the frequency response magnitude minus
20 dB,

max(𝐻 ̃)
𝜖= .
10
𝜖
𝛼(𝜔) = max ( ̃(𝜔).
− 1,0) 𝐻
̃
𝐻 (𝜔)

When 𝐻 ̃(𝜔) is larger than the threshold, then 𝛼(𝜔) = 0 (no regularisation). Conversely, when
𝐻̃(𝜔) < 𝜖, then 𝛼(𝜔) has a strictly positive value, which prevents the denominator to be too close
to 0.

Calculation of the inverse filter in the time domain

So far, the inversion has been carried out in the frequency domain, so an inverse Fourier transform
(IFFT) must be performed to obtain the inverse filter in the time domain. A typical characteristic of
the inverse filter calculated with regularisation is that it is non-causal, i.e., the filter starts responding
at negative times. This behaviour is managed by using a minimum-phase filter that has the same

244
magnitude response as the inverse filter, but which does not aim at correcting the phase response
of the device (Norcross, et al., 2004).

Interface of the inverse filter tool

The interface contains 6 different graphs (presented in chronological order):

1. The device’s impulse response in time domain, as measured and stored in a wave file. Note
that for headphones, these impulse responses should be relatively short.
2. The device’s frequency response in frequency domain, obtained from Graph 1 by FFT.
3. The calculated inverse filter frequency response, obtained by regularised inversion of Graph
2.
4. The inverse filter’s impulse response in the time domain, as it will be saved as a wave file.
It is obtained using the minimum phase filter algorithm on Graph 3.
5. Result impulse response shows the convolution between Graph 1 and Graph 4, for
verification. Ideally, this graph should be as close as possible to a Dirac function.
6. Result frequency response shows the product between Graph 2 and Graph 3, for verification.
Ideally, it should be a flat line equal to 0 dB. Narrow notches are acceptable for auralisation
purposes.

Overall, the process is fully automatic, but it can still be adjusted via two parameters, namely the
onset time and the filter length. Their values are suggested automatically (either if you open the
device’s impulse response for the first time, or if you click on the Automatic settings button).

The onset time indicates the start of the impulse response. In the inverse filter calculation, we
discard the signal before the onset time. The onset time is displayed on the impulse response graph
(Graph 1) as a black vertical line. It is automatically detected at the point where the amplitude is

245
40 dB lower than the peak. The user can modify the onset time by clicking and dragging the black
line. Note that it is not possible to place it after the impulse response peak.

The filter length is the length of the inverse filter. The device’s impulse response is truncated to
this parameter before applying the FFT and calculating the inverse filter. It is a power of 2 for an
optimal use of the FFT algorithm. The parameter is chosen in the corresponding drop-down list at
the bottom of the window (possible filter length values between 1024 and 32768). The filter should
be applicable to frequencies down to 20 Hz, which means it should be at least 1/(2*20 Hz)=1/40 Hz,
i.e. 25 ms long. Depending on the sampling rate, the proposed filter length is the next power of 2
in number of samples that corresponds to this duration. A longer filter is more accurate, but it is
also slower.

Note: for typical sampling frequencies of 44.1 kHz and 48 kHz, the automatic filter length is 2048
samples.

When closing the window, these two parameters are saved in a *.ini file. The next time the tool is
opened for that particular device impulse response, the onset time and the filter length are taken
from the *.ini file instead of being calculated automatically. If the ini file exists, it is always possible
to reload it with the appropriate button at the bottom of the window.

Note: this ini file is not compatible with versions earlier than ODEON 16.

Remarks on multichannel files

The tool can be used for a two-channel impulse response (e.g., the two channels of a pair of
headphones). In that case, the same parameters (onset time, filter length) are applied to both
channels.

During the automatic detection process, the earliest onset is chosen as the onset time for both
channel

For a multichannel impulse response, you can select which channel is displayed in the dropdown
list at the bottom of the window.

246
Appendix E: Head Related Transfer
Functions (HRTF’s) and digital
filtering
To create binaural simulations, a set of HRTF's is needed. The HRTF's are different from subject to
subject. In principle, and for best accuracy, you can measure your own and import them into
ODEON using the Tools>Create filtered HRTF menu entry in the ODEON program. Measuring HRTF's
are, however a complicated and costly task, so using the supplied ones is the simplest option.

If you should be interested in creating your own sets of HRTF’s for ODEON, additional information
can be found when clicking F1 when inside the ODEON program. Here search for HRTF.

The imported HRTF's to use for auralisation are pre-filtered into octave bands in order to reduce
calculation time. The octave band filter parameters for the selected filter bank can be seen on the
filter bank name at the Auralisation Setup menu. The filter parameters are:

M
The M value is taken into account if the “Apply enhancement” option is checked. If the file name
contains the ‘word’ Mddd where ddd is a floating-point number, then localization enhancement was
applied to the HRTF's (Zang, et al., 1998). This means that frequency dips and notches in the
individual HRTF's have been exaggerated in order to improve the directional cues in the HRTF's.
The M-factor determines how much the dips and notches have been exaggerated; If M is 0 then the
effect is neutral, a value of 3.0 improves localization without too much undesired colouration. An
M value greater than 3.0 does not seem to give any noticeable advantages whereas a value less than
3.0 gives less colouration. The enhancement algorithms are further developments of those used in
earlier versions of ODEON, as a result the M factor can be set as high as 3.0, allowing a significant
improvement of the 3D experience through headphones without noticeable drawbacks.

A
When an A is appended after the M-factor, this tells that Minimize colouration effects (diffuse field approach)
was not enabled so the HRTFs were optimized for Anechoic conditions (no reflections present). If
no A is found in this place in the name then ODEON has attempted to minimize colouration effects;
the enhancements applied may give some colouration effects because some frequencies are
amplified for individual directions. When this option is checked, ODEON will try to accomplish
that colouration is kept at a minimum in a multi-reflection environment, that is if the average
frequency response for all directions in the set of HRTF's is kept neutral.

Sample rate
The sample rate of the HRTF. This sample rate should be the same as the sample rate of the signal
files (anechoic recordings) to be used. The supplied HRTF's are sampled at 44100 Hz.

Apass
Ripple of octave band filters in dB. Smaller is better, 0.5 dB is sufficient.

Astop
Maximum possible attenuation of octave bands. To allow complete attenuation of all reflections of
a 16 bit signal (96 dB dynamic range), Astop should be 96 dB, however due to auditory masking we

247
are not able to hear such differences so 40 dB is probably sufficient. Smaller Astop leads to shorter
calculation time (of the BRIRs).

Band overlap in percent

Octave bands implemented using FIR filters are not completely rectangular, it takes some
frequency span before they attenuate completely. An overlap between the filters of 100 percent
gives a smooth transition between the filters, which is probably a more realistic representation of
real world reflections than with shorter overlaps. At the same time, long overlap gives shorter
calculation time (of the BRIRs).

If you should need to use filters with other filter parameters e.g. Astop being 96 dB you should create
a filtered set of HRTF's with these parameters, use the File|Create filtered HRTF's option. Then from
within your room, select the new filter bank from the Auralisation Setup.

The ODEON installation includes some HRTF’s: the Kemar (Gardner, et al., 1994), which has to be
manually imported from the ODEON folder, and Subject_021Res10deg (Algazi, 2001) which is
already imported. If you need HRTF’s other than these ones, you should create a text file following
the same format as used in the files Unity.ascii_hrtf, Kemar.ascii_hrtf and
Subject_021Res10deg.ascii_hrtf. These files can be found in the \HeadAndPhone\EE\ and
\HeadAndPhone\EC\ directories. To import a set of HRTF’s, select the Tools|Create filtered HRTF’s option,
then select the specific HRTF ASCII file e.g. C:\ODEON…\HeadAndPhones\EE\Kemar.ascii_hrtf, finally specify
miscellaneous parameters when the import dialog appears – for help on these parameters please
press F1 from within the dialog. You may also desire to import a set of HRTF’s which has already
been imported in order to specify an alternative filtering approach e.g. to enhance the HRTF’s in
different ways.

Headphone filters and HRTF pairing

It is possible to compensate for non-linear frequency response of headphones. When a headphone
is selected, ODEON will filter the audio output with a minimum phase filter, which has a frequency
response inverse to that of the headphones.

A number of headphone filters are supplied in ODEON's .hph format. Filters ending on .ee.hph are
measured on a dummy head at the entrance of a blocked ear canal whereas filters ending on .ec.hph
are measured at the end of the ear canal (at the ear drum, so to speak). The selected headphone
filter should match the HRTF’s used; for example, the Subject_021Res10deg.hrtf doesn’t have an ear
canal, therefore an .ee filter should be used, whereas the Kemar.hrtf does have an ear canal, so the .ec
filters should be used. If the corresponding filter for the headphone used is not available then a
generic filter matching the set of HRTF's may be used, e.g. Subject_021Res10deg_diffuse.hph. If a diffuse
field filter (equalization) is selected, the results are filtered in order to obtain an overall flat
frequency response of the HRTF's; that is, the average frequency response of all the HRTF filters
are calculated and the auralisation results are filtered with the inverse of that. If using headphones
which are diffuse field equalized (most headphones attempt to be) and a matching headphone filter
is not available, then the matching diffuse filter headphone filter can be used.

For your convenience, two directories with matching HRTF’s and headphone filters are installed
with ODEON - the \HeadsAndPhones\EE\ and \HeadsAndPhones\EC\ directories. The first directory contains
a set of HRTF’s which were measured on a live subject with blocked ear-canal by the CIPIC
Interface Laboratory (Algazi, 2001) along with our matching .ee.hph headphone filters. The next
ones contains the well-known Kemar HRTF’s, which includes ear-canals along with our matching
.ec.hph filters. One of the directories can be selected in the Options>Program setup>Auralisation setup

248
dialog – once done - matching the HRTF’s and headphone filters can be selected in the same dialog,
as well as in the Auralisation setup in the individual room.
If installing new HRTF’s (or installing some of the additional CIPIC data which can be downloaded
from http://www.odeon.dk/hrtf), headphone filters are acceptable in the .wav format. In that case
the filters should contain the impulse response(s) of the headphone as measured on a dummy head
of the same type as the one selected in the HRTF drop down menu (i.e. with or without ear canal
or a corresponding ear coupler). The filter may be one or two channels; two channels are desirable
if compensating a specific headphone, which is preferable than compensating for the model (the
actual headphone may deviate slightly from the model specifications). The measurement of the
impulse response of headphones can be done using ODEON’s Measurement System (see Chapter
12), and creating the inverse filter to be used will be taken care of by ODEON.

249
Appendix F: Impulse Response
processing tools
(Auditorium and Combined editions only)

A group of tools makes it possible to perform a number of advanced operations on impulse

responses to enhance measurements and auralisations. Currently, three tools are available at the
Tools menu in the main ODEON bar.

Create average of multiple impulse responses (.wav)

You can load as many impulse responses as you want at once (use SHIFT/CTRL+select in the open
dialog) and create an average Impulse response out of them. The tool is particularly useful when
you want to measure the response of headphones (using an artificial ear or a dummy head) or the
response of a speaker at the same position. Since every time you measure there might be slight
differences in the response, it is recommended to take more than one measurement, to minimize
variations. Use settings described below.

Type of averaging
There are two ways to create the average:
• In Frequency-domain Average, where the Impulse responses are first converted to
Frequency responses using FFT, averaged and then converted back to an Impulse response
(time domain) with zero imaginary part and minimum phase. This is the default setting
yielding the most stable results, when there are many 'outliers' in the selected group of
responses. This average is preferred for headphone measurements.
• In Time-domain Average (also called ensemble average), where the Impulse responses are
averaged directly in the time-domain sample by sample. In most of the cases, both time-
domain and frequency-domain averaging yields the same result. This average is
particularly useful for suppressing noise.

Time window before peak

This setting helps detecting the peak of impulse each response successfully, so that they are
perfectly aligned during averaging. The length of the window gives enough length to include any
gradual build-up of the energy at the onset (just before the peak).

250
Calculate and save
When all files are selected and loaded, press Calculate and Save to create the average impulse
response.

Create impulse response file (.wav) from spectrum (.txt)

The tool uses up to two .txt files that contain the spectrum of a device (microphone, loudspeaker,
headphone) and creates a 1-channel or 2-channel impulse response file from it. The .txt file is

251
typically obtained from the manufacturer of the
device and should contain one column for the
frequency and one column from the dB level.
Such a .txt file is displayed on the right figure.
This particular file accompanies the USB
microphone UMIK, by MiniDSP, and corre-
sponds to the particular serial number of our
device. Therefore, it is fully representative of the
frequency response of the item. The file has a
total of 616 lines with frequencies from 10.054 to 20016.816 Hz. You don’t have to delete the first
information line, as ODEON will automatically identify where the actual data begin. The tool starts
the process, by interpolating a high-resolution spectrum (frequency response) from the data given
in the .txt file. The interpolated spectrum for the .txt file shown above is illustrated in the following
figure.

The high-resolution spectrum is the then converted to time-domain using Inverse FFT and
assuming zero imaginary part and zero phase.
Press ‘Create Impulse
Response’ as in the screenshot
on the right. The tool will
create an impulse response
which would give the same
spectrum as the original one.
This impulse response can be
loaded in the next tool to
obtain an inverse filter that can
be used during measurements
or auralisation, in order to
neutralize the frequency
response of the device.

252
Create inverse filter (.wav)
The tool loads an impulse response, calculates its FFT frequency response and then the inverse
spectrum. This inverse spectrum is finally converted back to a time-domain impulse response,
which can be used either in measurements or auralisations to remove any frequency colouration
from the device in use (see below in typical cases). The tool uses the calculation engine described
in Appendix D. The following steps should be followed for using the tool:
• Load an impulse response using the Load Impulse Response button.
• Press the Create Inverse Filter button to calculate the inverse filter out of the impulse response.
• ODEON will now open another window, showing the calculated inverse filter and the
corresponding impulse response.
• Close the window. You will be prompted to save the file.

The tool can process up to 2-channel files.

253
Typical cases for using the tools above
Although the tools can be used alone, they will be mostly useful when used in combination for the
following cases:

Headphone measurements for auralisations

Before making any auralisation, it is advisable to compensate for the frequency response of the
actual used Headphone in the Auralisation setup. This is because the headphone has to remain as
neutral as possible in the playback chain and do not colourize the sound file with each own
frequency response.
ODEON comes with some default headphone filters that can be used directly in the Auralisation
setup. However, if a particular model is not in the list, you can make measurements yourself, using
an artificial ear or a dummy head with microphones on its ears, together with ODEON's Sweep
Measurement Facilities.
• You should first obtain 3 to 5 impulse responses, by removing and repositioning the
headphone each time on the artifial ear or dummy head. This will ensure that you will
measure a variety of slightly different positions on the ear, as it happens in real-life.
• Use the Create average of multiple impulse responses (.wav) tool to obtain an average Impulse
Response from the previous measurements.
• Use the Create Inverse filter (.wav) tool to make an inverse filter for auralisations
• Alternatively, simply copy the averaged Impulse Response (.wav) file in ODEON'S
HeadAndPhones\ee folder and select it directly in the Auralisation setup.

Microphone inverse filter for room acoustic measurements

This is useful for neutralizing the response of a microphone during measurements with ODEON's
Sweep Measurement Facilities.
• Typically, the manufacturer of a professional microphone can supply a .txt file with the
frequency response in two columns: 1st column containing the frequencies and 2nd column
containing the response in dB.
• Use the .txt file to create the corresponding impulse response of the device with the Create
impulse response file (.wav) from spectrum (.txt) tool.
• Create the inverse filter of the Impulse response by using the Create inverse filter (.wav) tool.
• Once you save the inverse filter file, make sure to press the Copy inverse filter button to copy
the file in the Measurements filter folder.
• The inverse filter file should now be available under the Input divide settings in the ODEON's
Sweep Measurement Facilities. See below.

254
Since version 17, ODEON comes with an example of .txt spectrum file, an impulse response and
an inverse filter file, corresponding to UMIK, by MiniDSP microphone. They are all located in the
Measurements folder.

Caution! The UMIK spectrum and Inverse filter files that come with the ODEON installation are only
there for demonstration purposes, since they match a particular item, by the serial number. Therefore,
they will not be as accurate if you attempt to apply to another UMIK item. In principle, you will have to
obtain the .txt spectrum file that corresponds to your own device.

The exact same procedure can be followed for an output device as well, i.e. for a omni-directional
loudspeaker used in measurements.

255
Appendix G: Modelling rooms in the
ODEON Editor
The ODEON .Par modeling format /language
Geometry models can be made using the parametric modelling language, which is built-in
to ODEON. The model data is typed into a text file with the file extension .Par using the
modelling language described below. You may use the supplied editor ODEONEdit to
create and edit your text files. The ODEON modelling format is not case-sensitive, so upper and
lower-case letters can be used as desired.

A simple modelling example

At its simplest (but not fastest), a floor with the dimensions 4 x 4 metres can be defined as follows
(using the reserved keywords Pt and Surf in order to define points and surfaces):
Z

FloorSurface.Par
###
O

Pt 1 0 0 0 X
Y

Pt 2 4 0 0
Pt 3 4 4 0
Pt 4 0 4 0
Surf 1 floor
1 2 3 4
###

One may choose to model the room point-by-point and surface-by-surface as in the example above.
However, for many geometries, it will be an advantage to use parameters to describe basic
dimensions in the rooms and to use high level statements to describe multiple points and surfaces
in a fast and flexible way. Before starting your first large modelling project it is a very good idea to
read through Chapter 3 or at least skim it - it will pay off in the end.

Another way to learn about the modelling language is to study the examples, which are installed
in the \ODEON...\Rooms\Manual samples directory along with the ODEON program – open the room(s)
in ODEON, then click the Open the ODEON Editor icon on the toolbar in order to study the geometry
file.

Components in the modelling format

The basic function of the modelling format is to allow modelling of surfaces in room geometries.
The surfaces can be modelled point-by-point, surface-by-surface, however it is also possible to
make use of symmetry and to create repeated features in a room such as columns, using loops.
Finally, it is possible to use hybrid functions, which create points as well as surfaces in terms of
shapes such as boxes, cylinder and domes.

Constants, variables and counters

Constants and variables can be defined and used in the file format. It is a good habit to use
constants whenever a value is used more than a few times in a file, this reduces typing errors and
it also makes it easier to make general changes to geometry such as changing the height of a room.

Mathematical expressions
Mathematical expressions can be used to express any real or integer number in the file e.g.
coordinates, constants, variables, counters, point numbers, surface numbers, etc. If you use a value

256
that is not an integer, to describe a point or surface number, then that value will be rounded to the
nearest integer value.

You may describe coordinates using mathematical expressions like:

Length*Sin(PI/4)

where Length is a user-defined constant or variable. Mathematical expressions may not contain any
SPACE or TAB (tabulation) characters. To get a complete overview of the mathematical functions
available, please refer to appendix F.

Points
A point is made up from a unique point number and its X,Y and Z coordinates. Use the Pt, MPt and
CountPt statements to define points. Points can also be defined implicitly, using one of the hybrid
statements.

Surfaces
A surface is made up from a unique number, an optional descriptive text and a number of points
connected to one another. All the points must be coplanar; otherwise the surface cannot be
constructed. To define surfaces, use the Surf, MSurf, CountSurf, ElevSurf, ElevSurf2 and RevSurf
statements. Surfaces may also be defined implicitly using the hybrid statements.

Hybrid statements
Hybrid statements are: Box, Cylinder, Cylinder2, Cone, Dome, Dome2 and ElevSurf. The hybrid statements
create the points and surfaces needed to model the specified shape. The points and surfaces created
must always have unique numbers.

Coordinate manipulation functions

A set of functions for coordinate manipulation (and surfaces made up from coordinates) is
included. This includes rotation around the various axes, scaling and translation. These functions
are needed in order to insert shapes defined by the hybrid statements in the geometry with the
correct position and orientation.

Comments and empty lines

Lines containing comments, and empty lines, may be inserted anywhere in the file, as long as they
do not come between data items, which should occur on one line. Comment lines must begin with
a colon (:), a semicolon (;), a slash (/) or an asterisk (*). The semicolon can also terminate a non-
comment line, allowing the non-comment line to be followed by a comment.

A series comment lines are started with a ‘{‘ and terminated by a ‘}’ at any position on a line. This
is useful to disable a section of lines.
Syntax highlighting in the ODEON editor makes it easy to spot comment lines.

Reserved keywords, predefined counters and constants

The following keywords are reserved by ODEON and have a special meaning in the parametric
modelling language:

Line folding markers

BeginBlock, EndBlock

Constant and variable statements

Const, Var

257
Point statements
Pt, MPt, CountPt

Point list statements

Plist0 – Plist9
ResetPList0 – ResetPList9
PlistA, PListB

Surface statements
Surf, MSurf, RevSurf, CountSurf, ElevSurf, ElevSurf2

Hybrid statements
Box, Cylinder, Cylinder2, Cone, Dome, Dome2

Loop statements
For…End,

Transformation statements
Mreset, MPop, MScale, MTranslate, MRotateX, MRotateY, MRotateZ
(and for compatibility with earlier releases of ODEON: Scale, UCS)

Predefined constants
PI = 3,14159265358979312

Predefined variables
NumbOffSet, ONVert

Predefined Counters
PtCounter

Coordinate system definition statements

Unit, CoordSys

Debugging Facilities
DebugIsOn, Debug

Line folding markers

Line folding is a feature of the ODEON editor where a section of lines (e.g. a part of the geometry)
can be collapsed (folded) into one line in the editor for a better overview. This is the only
functionality of the two keywords BlockBegin..BlockEnd, they are ignored by ODEON when a .par
file is loaded. The keywords are automatically generated when a room is generated by
ODEONExtrusionModeller or when a room is imported in the .dxf format.

BeginBlock <optional comment>

NumbOffSet 100
Pt 1 0 -1 0
Pt 2 0 1 0
Pt 3 1 1 0
Pt 4 1 -1 0

Surf 1 A surface
1 2 3
EndBlock <optional comment>

258
Any code between a corresponding set of BlockBegin and BlockEnd lines can be collapsed by
clicking a small + in the left side of the editor, making it easier to handle large files, in particular
with many layers. All blocks can be collapsed from the view menu in the ODEONEditor. And
Blocks can be nested, that is, have Blocks contained within Blocks.

Defining constants
Constants must follow the syntax:

Const <Name> <Value>

where <Value> is a mathematical expression, which may be based on numbers or constants and
variables that have already been defined.

Example 1:

Const CeilingHeight 3.4

Example 2:

Const FloorLevel 1
Const CeilingHeight FloorLevel+3

Example 3:

Const FloorLevel 1
Const Length 6
Const CeilingHeight FloorLevel+Length*TanD(30)

Defining and reassigning variables

The definition of variables must follow the syntax:

Var <Name> <OptionalValue>

Example 1; defining the variable FloorLevel:

Var FloorLevel

Example 2; defining the variable FloorLevel and assigning the initial value 0:

Var FloorLevel 0

Example 3; reassigning a variable /adding 1 metre to the FloorLevel:

FloorLevel FloorLevel+1

Remark: The predefined variable NumbOffSet may be used like any other variable, but has a special
meaning because it offsets point and surface numbering. This variable is useful if copying a part of
a geometry from another geometry file, it is also useful in connection with the for..end statements.
Auto can also be assigned to NumbOffSet, in doing so ODEON will automatically increment the value
of NumbOffSet to be greater than any point and surface number previously defined. This has the
advantage that repeated point and surface numbers can easily be avoided without having to keep
track on the numbers used - the drawback is that slight changes in the geometry file may change
numbers on many subsequent surfaces, ruining the relationship between surface numbers and the
material assigned to that surface inside the ODEON program. The Auto option is very useful in

259
combination with loop constructions (see description of the for..end constructs later on). Typing
PtAbsRef after the value assigned to NumbOffSet forces absolute number references for points while
using the specified offset on the numbers of surfaces – this is explained later.

Example on the use of NumbOffSet; creating surface 101 containing the points 101 to 104 and surface
201 containing the points 201 to 204:

NumbOffSet.Par
### Z

NumbOffSet 100
Pt 1 0 -1 0
Pt 2 0 1 0
Pt 3 1 1 0
Pt 4 1 -1 0

Surf 1 A surface
O
1 2 3 4

NumbOffSet NumbOffSet+100 Y

Pt 1 0 -1 1
Pt 2 0 1 1 X

Pt 3 1 1 1
Pt 4 1 -1 1
Surf 1 Another surface
1 2 3 4
###

Example creating points 1–4 and surface 1, setting NumbOffSet to Auto, then creating Point 5-8 and
surface 5:
###
Pt 1 0 -1 0
Pt 2 0 1 0
Pt 3 1 1 0
Pt 4 1 -1 0

Surf 1 A surface
1 2 3 4

NumbOffSet Auto
Pt 1 0 -1 1
Pt 2 0 1 1
Pt 3 1 1 1
Pt 4 1 -1 1
Surf 1 Another surface
1 2 3 4
###

Defining a point using the Pt statement

Use the Pt statement to define a single point. The syntax must be as follows:

Pt <Point Number> <XMathExpression> <YMathExpression> <ZMathExpression>

Example defining point number 100 in (x,y,z) = (1,1,1):

Pt 100 1 1 1

Hint: Point number and coordinates can be written using mathematical expressions, allowing
greater flexibility and reusability.

260
Parametric modelling – defining multiple points
Use the MPt statement to define a series of points, which is typically used in connection with the
ElevSurf or ElevSurf2 Statement. The syntax must be as follows:

MPt <Number> <NumberOfPoints>

<XMathExpression1> <YMathExpression1> <ZMathExpression1>
<XMathExpression2> <YMathExpression2> <ZMathExpression2>
....NumberOfPoints lines each defining a point in the multi point sequence, should follow the MPt statement.

<Number>
A unique number from 1 to 2.147.483.647 for identification of the first point in that multi point
sequence.

<NumberOfPoints>
The number of points defined by this multipoint statement - if the number is 3, then 3 lines should
follow, each describing the coordinates of a point.

Example 1; defining point number 100 in (x,y,z) = (1,1,1) and point number 101 in (x,y,z) = (2,2,2)

MPt 100 2
1.0 1.0 1.0
2.0 2.0 2.0

As a special option for multi points, it is possible to repeat a coordinate used in the previous point
of that multipoint sequence or to repeat the coordinate while adding or subtracting a value from
that point:

Example 2; defining point number 100 in (x,y,z) = (1,1,1) and point number 101 in (x,y,z) = (1,2,0)

MPt 100 2
1 1 1
= =+1 =-1

Defining a series of points using the CountPt statement

The CountPt statement must follow the syntax:

CountPt<FirstPointNo><MaxCount><XMathExpression><YMathExpression><ZMathExpression>

Use the CountPt statement to define a series of points using a counter. This statement makes use of
the predefined counter PtCounter, which will run from 0 to MaxCount-1, producing the points with
the numbers FirstPointNo to FirstPointNo+MaxCount-1. Use the PtCounter in the expression of the x,y
and z coordinates to create the desired differences between the ‘count points’.

Example; defining 7 points on a circle with a radius of 10 at Z=0 metres:

CountPt 100 6+1 10*CosD((PtCounter)*360/6) 10*SinD((PtCounter)*360/6)

0

Note! First and last point in this series of ‘count points’ are equal (redundant). This will typically
be desirable when using the CountPt statement along with RevSurf statement.

Defining a single surface using the Surf statement

A Surf statement is divided into two lines and must follow the syntax:

Surf <SurfaceNumber> <Optional Description>

<ListOfPointNumbers>

261
The Surf statement is used to define a single surface (in some situations with symmetry, two
surfaces). The Surf statement is constructed from two lines, one identifying the surface by a number
and an optional name and another with a list of corner numbers.

<SurfaceNumber>
A unique number from 1 to 2,147,483,647 for identification of the surface. Using the same number,
but with negative sign defines the surface and its counterpart mirrored in the XZ-plane (Y = 0). The
surface number may be defined using mathematical expressions.

<Optional Description>
A string displayed and printed for easy identification of the surface, i.e.: ‘Main floor'.

<ListOfPointNumbers>
Each surface may be bounded by between 3 and 500 corners, which all lie in a plane. Corner
numbers refer to the corners, which must have been defined (e.g. using the Pt or CountPt statements)
before using the surface statement. The order of listing must be as obtained by travelling around
the surface's perimeter (in either direction). The list of corner point numbers must be on the same
line. A room may contain up to 10000 surfaces.

Example 1, surface made from point 1, 2, 3, 4:

Surf 100 floor

1 2 3 4

Example 2, surface made from point 1,2,10,11,12,13,14,4,5:

Surf 200 Ceiling
1 2 10>14 4 5
If there is a need to programmatically build a list of points, this can be done using the PList and
ResetPList statements.

Building lists of points using PList and ResetPList

The PList and ResetPList statements are used in special cases together with the Surf statement. Twelve
lists are predefined namely PList0 to PList9 (and PlistA and PListB which are handled automatically
by ODEON when modelling a box and other solids). The PList statements allow to
programmatically construct a list of points e.g. a list like:

100 110 120 130 140 150 160 170 180 190 200

this can be done using a for..end construct in the following way (adding a point number at a time):

for MyCounter 0 10
PList0 100+MyCounter*10
end

It is also possible to add a number of points to a point list, e.g. another PList to a PList. In the
following example PList1 is assigned the points 100 110 120 130 140 150 160 170 180 190 200 10 11
12 13 15):

Plist1 Plist0 10>13 15

A point list can be referenced in the following way (adding point 1 before and 2 after the list in this
example):

262
 Surf Test_surface
1 PList0 2

To reset the list use the statement (list 0 used in this example):
ResetPList0

Multi Surface - MSurf

The multi surface MSurf is essentially just a variant of the Surf statement. Instead of typing one
header line (e.g. Surf 1 A surface name) for each surface, the header can be shared by multiple surfaces.

MSurf <SurfaceNumber> <NumberOfSurfaces> <Optional Description>

<ListOfPointNumbers1>
<ListOfPointNumbers2>
<ListOfPointNumbers3>
<NumberOfSurfaces> lines with lists of points describing each surface.

<SurfaceNumber>
A unique number from 1 to 2,147,483,647 for identification of the surface. Using the same number,
but with a negative sign defines the surface and its counterpart mirrored in the XZ-plane (Y = 0).
The surface number may be defined using mathematical expressions.

<NumberOfSurfaces>
The number of surfaces in the MSurf.

<Optional Description>
A string displayed and printed for easy identification of the surfaces. Could be something like
Main floor.

Example on multi surface, containing 5 sub-surfaces:

MSurf 1 5 Steps on a stair
5544>5534 5112>5122
5111>5101 5212>5222
5211>5201 5312>5322
5311>5301 5412>5422
5411>5401 5512>5522

Elevation surface Z

Use the ElevSurf statement to define a series of vertical surfaces O

Y
X
from a series of perimeter points plus an elevation height. The
perimeter points are typically defined using the MPt statement.
The syntax of ElevSurf is:

ElevSurf <FirstSurfaceNumber> <FirstPointNumber> <SectionsInElevSurf> <Height> <Optional name>

Example on use of the MPt and ElevSurf statements:

First the perimeter points (point 1 to 23) at the floor level of an office environment are described
using the MPt statement. Then the elevation surface is created from these points, creating the
perimeter walls of the office with a constant height of 2.7 metres. Finally, the floor and ceiling are
created using the Surf statement.

263
 0.00 10.00 20.00 30.00 40.00 metres

18 19 22 1

14 17 20 21
12 15 16 7 6
13 5 4
10 11
3 2
9 8

Perimeter points at the floor level

Example file: MPt and ElevSurf.Par

Demonstrates the use of MPt (multi point), Surf (Surface) and ElevSurf (Elevation surface) statements.
In this example the X-coordinates are expressed in absolute values whereas the Y-coordinates in
most cases are increased or decreased using the =+ or =- operators, respectively.

To create a closed ElevSurf (that is, the first wall joins the last wall), the first and last point in the
series of points handled to the ElevSurf must be identical - in this example, point 1 and point 23 are
identical. If an elevation surface has 22 surfaces, then 23 points must be made available to the
ElevSurf as in this example.
###
MPt 1 23
000
= 7.48 =
2.3 = =
= =-2.38 =
8.35 = =
= =-1.26 =
10.76 = =
= =+4.64 =
64.78 = =
= =-2.68 =
51.52 = =
49.62 =-2.02 =
48.22 =+1.1 =
45.3 =-2.68 =
43.85 =+1.45 =
42.40 = =
40.98 =-1.45 =
=0=
34.5 = =
= =2.1 =
30.13 = =
= =-2.13 =
000

ElevSurf 1 1 22 2.7 walls

Surf 200 Floor

1>22

Surf 201 Ceiling

24>24+22-1
###

Elevation surface 2
Use the ElevSurf2 statement to define a series of vertical surfaces from a series of perimeter points
plus an elevation height. The perimeter points are typically defined using the MPt statement. The
syntax of ElevSurf2 is:

264
 ElevSurf2 <FirstSurfaceNumber> <FirstPointNumber> <SectionsInElevSurf> <Height> <T/B/N> <Optional
name>

The ElevSurf2 only differs from ElevSurf in that a top and bottom surface may be specified (the
T/B/N option).

<FirstSurfaceNumber>

A unique number from 1 to 2,147,483,647 for identification of the first surface in the ElevSurf2
surface. Using the same number, but with negative sign, defines the surfaces and their counter
parts mirrored in the XZ-plane (Y = 0).

<FirstPointNumber>
First point number in the floor perimeter. The floor points are typically MPt statement. See example
below.

<SectionsInElevSurf>
The number of surfaces to be created by the ElevSurf2 statement. If creating a cylinder, a number
between 16 and 24 is suggested (if it’s a column, use only six to eight surfaces).

<Height>
Height is oriented in the Z-direction.

<T/B/N>
The T/B/N parameter specifies whether the ElevSurf2 should have a top and /or a bottom. The
options are T, B, TB and N (for none). Top or bottom may only be included if all of the points in the
floor in the elevation surface are in the same plane.

Example on use of the MPt and ElevSurf2 statements.

First the perimeter points (points 1 to 23) at the floor level of an office environment is described
using the MPt statement. Then the elevation surface is created from these points, creating the
perimeter walls of the office with a constant height of 2.7 metres.

In this example the X-coordinates are made in absolute values whereas the Y-coordinates in most
cases are increased or decreased using the =+ or =- operators, respectively.

To create a closed ElevSurf2 (that is, the first wall joins the last wall), the first and last point in the
series of points handled to the ElevSurf2 must be identical - in this example, point 1 and point 23
are identical.

If an elevation surface has 22 surfaces, then 23 points must be made available to the ElevSurf2 as in
this example.

###
MPt 1 23
000
= 7.48 =
2.3 = =
= =-2.38 =
8.35 = =
= =-1.26 =
10.76 = =
= =+4.64 =
64.78 = =
= =-2.68 =
51.52 = =
49.62 =-2.02 =

265
 48.22 =+1.1 =
45.3 =-2.68 =
43.85 =+1.45 =
42.40 = =
40.98 =-1.45 =
=0=
34.5 = =
= =2.1 =
30.13 = =
= =-2.13 =
000

ElevSurf2 1 1 22 2.7 TB walls

###

Defining a number of surfaces using the CountSurf statement

The CountSurf is mostly available for backwards compatibility. In most cases it will be easier to use
the Surf statement along with a for….end loop.
A Counter surface is divided on two lines and must follow the syntax:
CountSurf <First Surface Number> <NumberOfSurfaces> <Optional name>
<ListOfPointNumbers>

<FirstSurfaceNumber>
A unique number from 1 to 2,147,483,647 for identification of the surface. Using the same number,
but with negative sign defines the surface and its counterpart mirrored in the XZ-plane (Y = 0). A
CountSurf will take up several surface numbers, which must all be unique.

<NumberOfSurfaces>
The number of surfaces to be created by the CountSurf call.

<Optional name>
Optional user defined name for easy identification of the surface, e.g. ‘Beam'.

<ListOfPointNumbers>
Each surface may be bounded by between 3 and 50 corners, which all lie in a plane. Point numbers
refer to the corners, which must have been defined (e.g. using the Pt or CountPt statement) before
using the surface statement. The order of listing must be as obtained by travelling around the
surface's parameter (in either direction). The list of corners must be on the same line of code. A
room may contain up to 10000 surfaces.

Example:

CountSurf 1000 5 Beam in ceiling

1000 1100 1200 1300

will produce five surfaces, the first one containing the numbers given in the ListOfPointNumbers, and
the next surface with 1 added to all the corners in the list, etc. Of course, all the points referred to
need to be defined, typically this is done using a CountPt definition for each of the corners referred
to in the corner list of the CountSurf statement. In the above example the points 1000-1004, 1100-
1104, 1200-1204 and 1300-1304 need to be defined.

Sample room files:

Beams.Par
BeamBox.Par
BeamBoxWithWindows.Par

266
Revolution surface RevSurf
RevSurf must follow the syntax:

RevSurf <FirstSurfaceNumber> <CurveStart1> <CurveStart2> <SectionsInRevSurf> <Optional name>

The RevSurf command is typical used together with two CountPt statements to create a revolution
surface using two ‘curves' of points. The curves must contain the same number of points. The
RevSurf command will always create a number of surfaces each build from four points.

<FirstSurfaceNumber>
A unique number from 1 to 2,147,483,647 for identification of first surface in the revolution surface.
Using the same number, but with negative sign, defines the surface and its counterpart mirrored
in the XZ-plane (Y = 0).

<CurveStart1>
First point number in the first revolution curve. The ‘curve of points' is typically created using the
CountPt statement and the curve must contain one more point than number of sections in the RevSurf.
<CurveStart2>
First point number in the second revolution curve. The curve of points is typically created using
the CountPt statement and the curve must contain one more point than the number of sections in
the RevSurf. The second curve must always contain the same number of points as the first curve.

<SectionsInRevSurf>
The number of surfaces to be created by the RevSurf statement. If creating a cylinder, a number
between 12 and 24 is suggested. Although it is easy to create many surfaces in a revolution surface,
too many small surfaces should be avoided. If the FirstSurfaceNumber is 100 and SectionsInRevSurf is
3, surfaces 100, 101 and 102 will be created.

<Optional name>
Optional user defined name for easy identification of the surface, e.g. ‘cylindric wall'.

Example:

RevSurf 1000 100 200 6 Cylinder

creates a revolution surface divided in 6 surfaces (surface 1000 - 1005). This call requires two curves
of each 6+1 points to be defined, namely point 100 to 106 and point 200 to 206. If the two curves of
points define corners in the lower and upper edge of a cylinder, a cylinder of 6 sections is created
(see example room: RevSurfCylinder.Par)

Loop surface <LSurf>

267
LSurf must follow the syntax:

LSurf <FirstSurfaceNumber><NumberOfLoops><Optional name>

<ListOfPointNumbers1>
<ListOfPointNumbers2>
<ListOfPointNumbers3>
<NumberOfSurfaces> lines with lists of points describing each loop - the first line preferably being the
outer loop

The LSurf command is typical used for creating walls with window holes in them. In the special
case where NLoops is equal to 1 the surface is identical to that created using the Surf command.

<FirstSurfaceNumber>
A unique number from 1 to 2,147,483,647 for identification of the first surface in the LSurf surface.
Using the same number, but with negative sign, defines the surface and its mirrored counterpart
in the XZ-pane (Y = 0).

<NumberOfLoops>
A least 1 for the outer perimeter of the surface. If e.g 3 then there is to addiotional loops
describing 2 holes (e.g. windows in the surface). In order to close the holes in the surfaces
additional surfaces has to be added e.g.using the Surf command.

<Optional name>
Optional user defined name for easy identification of the surface, e.g. ‘cylindrical wall’.

Example

Pt 1 0 0 0
Pt 2 5 0 0
Pt 3 5 0 3
Pt 4 0 0 3
Pt 5 2 0 1
Pt 6 3 0 1
Pt 7 3 0 2
Pt 8 2 0 2

LSurf 1 2 Wall with a window hole

1 2 3 4
5 6 7 8

... creates an LSurf with the outer perimeter defined by points 1-4, and a hole defined by points 5-
8.

Loops using the FOR....END construct

The For statement must follow the syntax:

For <CounterName> <CountFrom> <CountTo>

268
 <CounterName>
Name of counter to be used by the For statement. The counter is automatically defined by the For
statement and becomes undefined when the loop finishes. The counter can be referenced within
the for..end loop as an ordinary constant or variable if desired so.

<CountFrom>
First value the counter takes. The CountFrom value is considered an integer value. If the number
entered here is not an integer, it will be rounded to the nearest integer value.

<CountTo>
Last value the counter takes. The CountTo value must be greater or equal to the CountFrom value.
The For statement will take CountTo-CountFrom+1 loops. The CountTo value is considered an integer
value, if the number entered here is not an integer, it will be rounded to the nearest integer value.

The following example will produce the points 1 to 5 with the X-coordinates 5, 10, 15, 20 and 25
metres, while the counter (MyCounter) loops through the values 1 to 5:

For MyCounter 1 5
Pt MyCounter MyCounter*5 0 0
end

When using For..End constructs, it should be remembered that point and surface numbers must be
unique. This is easily obtained by incrementing the special variable NumbOffSet appropriately in
each loop. An example on this kind of numbering can be found in the sample file
BoxColumnRoom.Par where NumbOffSet is incremented by eight in each loop (each time a new
column, which contains 4 surfaces and 8 points, is created).

Sample room files:

ForRotunde.Par
BoxColumnRoom.Par

Unit
The Unit statement is used if you wish to model in a unit different from metres. The unit used in
the parametric file is by default assumed to be metres, however if you prefer to model in another
unit, this is possible using the Unit statement.

Example, modelling in Inches:

Unit Inches

You may choose your unit among the following predefined:

Metres, Centimetres, Millimetres, Inches, Feet and Yards

Or if you need a different unit, simply type the scaling factor from your unit into metres, e.g.
Unit Inches

corresponds to:
Unit 0.0254

The Unit statement may be used more than once in the same .par file:

###
Unit Inches
;e.g. imported model data in inches
.....model data.......
.....model data.......
Unit 0.57634
; model data measured on a paper drawing which appeared in an odd unit

269
 .....model data.......
.....model data.......
Unit Metres
; model data appended in the ODEON editor / modelling environment
;it is most practical to use metres as the unit when modelling in the ODEON
;environment - then coordinate values will be the same in the Editor as inside the 3DView in ODEON
.....model data.......
.....model data.......
###

CoordSys statement
The CoordSys statement is used if you wish to redefine the orientation or the coordinate system in
which the geometry was modelled. The statement is typically used if the geometry was ‘by
accident’ modelled in an orientation different from the one assumed by ODEON or if it was
imported from a CAD drawing where the orientation may also be different. To obtain the easiest
operation inside ODEON the following orientation should be used (using a concert hall as the
example):

• X-axis pointing towards the audience Z

• Y-axis pointing to the right as seen from the O

Y
audience X

• Z-axis pointing upwards

The syntax is:

CoordSys <X> <Y> <Z>

where X, Y, Z indicate which axis should be used as the x, y and z axis inside
ODEON. X, Y and Z may also have a sign to indicate that the axis should point in the opposite
direction.

Example 1, the default orientation (which is assumed by ODEON if the CoordSys statement is not
used in the geometry file):

CoordSys X Y Z

Example 2, changing the direction of the X-axis:

CoordSys -X Y Z

Example 3, Swapping the X and the Z-axis

CoordSys Z Y X

Example 4, The CoordSys statement may be used more than once in the same .par file:

###
CoordSys -X Y Z
;e.g. if the X axis was inverted in imported model data
.....model data.......
.....model data.......
;resetting the coordinate system to the default
CoordSys X Y Z
; model data appended in the ODEON editor / modelling environment
;it is most practical if using the default Coordinate system when modelling in the ODEON ;environment -
then coordinate system will have the same orientation in the ODEON editor as well as in the 3DView inside
ODEON.
.....model data.......
.....model data.......
###

270
User coordinate system
The UCS command is mostly there for compatibility with previous versions of ODEON – the
coordinate manipulation functions MTranslate, MRotateX, MRotateY, MRotateZ, MScale, MPop and MReset
included from version 4.21 allow far more flexibility. For your own convenience, it is not advisable
to mix the UCS method with the M-family method.
The UCS command must follow the syntax:

UCS <TranslateX><TranslateY><TranslateZ><RotateZ>

The UCS command is used to create a User Coordinate System, with its own X, Y and Z translation.
It also allows a rotation around the Z-axis (specified in degrees). All point definitions made after a
UCS call will be created in the specified coordinate system. The default coordinate system is defined
as:

UCS 1 1 1 0

The UCS command corresponds to:

MReset
MTranslate <TranslateX><TranslateY><TranslateZ>
MRotateZ <RotateZ>

If the UCS command doesn’t fulfill your needs for coordinate manipulation, you may use the matrix
manipulation family MTranslate, MRotateX, MRotateY, MRotateZ, MScale, MPop and MReset.

Scale
The Scale command is mostly there for compatibility with previous versions of ODEON – the
coordinate manipulation functions MTranslate, MRotateX, MRotateY, MRotateZ, MScale, MPop and MReset
included from version 4.21 allow far more flexibility. For your own convenience, it is not advisable
to mix the Scale method with the M-family method.

The Scale command must follow the syntax:

Scale <ScaleX><ScaleY><ScaleZ>

The scale command will multiply/scale all the points generated after the scale call, using the
specified x, y and z-scale. The default setting is:

Scale 1 1 1

The Scale command evokes scaling of coordinates after all other coordinate manipulation is carried
out. If you require more advanced scaling options, you may use the MScale option.

Coordinate manipulations – the M-family

Advanced coordinate manipulation can be carried out using matrix manipulation. The coordinate
manipulation functions, which is essential to the use of hybrid statements ( Box, Cylinder, etc.) is
implemented as the following functions:

MTranslate <TranslateX><TranslateY><TranslateZ>
MRotateX <Rotation angle>
MRotateY <Rotation angle>
MRotateZ <Rotation angle>

271
 MScale <ScaleX><ScaleY><ScaleZ>
MPop
MReset

The manipulations carried out by the M-family are cumulative. This means that you can specify
more than one operation to be carried out, e.g. first rotate 90° around the Z-axis,
then rotate 90° around the Y-axis and finally translate 10 metres
upwards. The following example shows these operations carried out
on a cylinder shell (the Cylinder statement is described later):

Manipulating a cylinder.Par
### Z

MRotateZ 90
O
MRotateY 90 X
Y

MTranslate 0 0 10
Cylinder 1 20 5 180 10 TB Cylindrical ceiling
###

The transformation commands to be carried out must always be stated before the points/geometry
on which they should work is created. To reset all previous coordinate manipulations, use the
MReset command. To reset/cancel the most recent manipulation (MTranslate in the example above),
use the MPop command (which will pop the operation of the matrix stack).

Hints!
The order in which the coordinate manipulations are carried out is important: usually (but not
necessarily always), the MScale commands should come first, then the MRotate commands and
finally the MTranslate commands.
If you are not familiar with coordinate manipulations, it may be a good exercise to try different
manipulations on the sample geometry above and load the geometry into ODEON upon each
change.

Using layers in ODEON

The Layer statement allows dividing Z

geometry into separate parts, which can O

Y
be displayed separately and in its own X

layer-colour in the 3DView, 3DOpenGL

and Materials list. This makes it easier to
model and investigate selected groups of
surfaces. When importing geometry
from a .dxf file (e.g. from AutoCAD
where layers are an integrated part) the
layers included in that file will be
preserved in the imported version of the
room.

If layers have been used in geometry, the layer can be activated or deactivated in the 3DView,
3DOpenGL and Materials list. The layers menu is activated from these windows using the Ctrl+L
shortcut.

Vocabulary – what’s a layer?

Layers are commonly used in CAD modelling programs such as AutoCAD in order to make
complicated geometries more easily manageable. Layers in CAD programs (and some drawing and
picture editing programs) can be compared to overhead sheets (without any thickness). You define
a number of layers with different names (and possibly different line colour/thickness etc.) and draw

272
the different parts of your geometry on the different layers. The layers can be turned on or off in
the CAD program allowing better overview by hiding parts of the geometry that are not relevant
in a part of the 'drawing' process.

Syntax for the Layer statement:

Layer <"Layer name in quotes"> <R-intensity><G-intensity><B-intensity>

or as another option:

Layer <"Layer name in quotes"> <LayerColour>

<"Layer name in quotes">

A name which must start and end with a quote sign (").

<R-intensity><G-intensity><B-intensity>
Three floating-point values between 0 and 1, which together is describing the colour of the layer as
a Red-Green-Blue intensity. If using the Layer command Shift+Ctrl+L from within the ODEON
editor, the colour intensities are set by clicking the desired colour in a dialog-box. Do note that it’s
not advisable to choose a greyish colour as it may not be easily visible in ODEON.

<LayerColour>
As another option the colour of the layer can be described, using one of the predefined colours
Black, Blue, Cream, Fuchsia, Gray, Green, Lime, Maroon, Navy, Olive, Purple, SkyBlue, Teal, MoneyGreen or White.
The LayerColours.par example demonstrates the different colours.

The LayerStatement.par example shows how to create a geometry on three different layers.

Selected surfaces can be selected for display in the 3DView, 3DOpenGL and the Materials list.

LayerStatementRoom.Par
###
const L 40
const W 30
const H 3
const NumColX 4
const NumColY 3
const ColumnW 0.3

MTranslate l/2 0 0
Layer "Walls" 1.000 0.502 0.000 ;orange colour
Box 1 l w h tb walls in the room
MPop
:modelling the columns

for ColYCnt 1 NumColY

for ColXCnt 1 NumColX
MReset
MTranslate ColXCnt*L/(NumColX+1) w/2-ColYCnt*W/(NumColY+1) 0

NumbOffSet Auto
Layer "Columns" 0.502 0.502 0.000
Box 1 ColumnW ColumnW h n columns in the room ;olive colour
NumbOffSet Auto
MTranslate 0 0 1.2
Layer "Table plates" 0.000 0.502 1.000 ;bluish colour
Box 1 3 3 0.1 tb tables
end
end
###

273
Symmetric modelling
Symmetric rooms can be modelled, taking advantage of the ODEON convention for symmetric
models. This allows generation of symmetric or semi-symmetric rooms with symmetry around the
XZ-plane, (Y = 0) – symmetric modelling is always carried out in the main coordinate system, it
does not take into account manipulations carried out using UCS and MTranslate. Modelling a surface
as symmetric around the main axis' (e.g. a reflector above the stage) can be done using symmetric
points. Modelling left and right walls at the same time can be done using a symmetric double
surface.

Symmetric points
Surfaces symmetric around the XZ-plane, Y= 0 can be made using symmetric points. If defining the
point:

Pt 2 1.0 1.0 1.0

in the geometry file, using the point -2 in a surface definition of the geometry file will refer to the
auto generated point:

Pt -2 1.0 -1.0 1.0

Thus the following surface definition:

Surf 1000 Symmetric surface

1 2 -2 -1

will model a surface symmetric around the XZ-plane, Y=0 (e.g. an end wall or a reflector).

If the surface is completely symmetric as above, then the symmetric points can also be specified
using the Mirror word, which should be the last component in the corner list:

Surf 1000 Symmetric surface

1 2 Mirror

Note: You should not try to define the point -2 in the geometry file, as it is automatically generated.

Symmetric double surface

Symmetric double surfaces are pairs of surfaces symmetric around the XZ-plane, Y= 0 (e.g. a right
and a left wall):

Surf -2 Right wall/ Left wall

12 22 23 13

will appear as two surfaces inside the ODEON program. Thus, you will have the following two
surfaces (inside the ODEON program):

-2 Right/ Left wall

containing the symmetric points:

-12 -22 -23 -13

and the surface:

2 Right/ Left wall

274
containing the points:

12 22 23 13

as they are defined in the geometry file.

Note: You may not define surface 2 if you are using the symmetric double surface -2, because
ODEON automatically generates surface number 2.

The Box statement

The Box statement defines a Box with or without top and
bottom. The Box statement may typically be used for Box-
shaped rooms and columns. A special case of the Box
statement is when one of the dimensions Length, Width or
Height is zero; in this case only one surface is created. Z

The syntax of the Box statement is: O

Y
X

Box <Number> <Length> <Width> <Height> <T/B/N>

<optional name>

<Number>

A unique number from 1 to 2,147,483,647 for identification of the first point and surface in the Box.
Using the same number, but with negative sign defines the box and its counterpart mirrored in the
XZ-plane (Y = 0). A Box will take up several point- and surface numbers, which must all be unique.

<Length>
Length is oriented in the X-direction on the Figure.

<Width>
Width is oriented in the Y-direction on the Figure.

<Height>
Height is oriented in the Z-direction on the Figure.

<T/B/N>
The T/B/N parameter specifies whether the Box should have a top and /or a bottom. The options are
T, B, TB and N (for none).

Insertion point:
The insertion point of the Box is always the centre of the floor (bottom) surface.

Connection points:
The four foot-points in Box are stored in PListA
The four top-points in Box are stored in PListB

The Box example shown was generated with the following code:

BoxStatement.par
###
const L 6

275
 const W 4
const H 2.7

Box 1 L W H TB Walls, floor and ceiling

###

The Cylinder statement

The Cylinder statement defines a cylinder shell with or without
top and bottom. The statement may typically be used for
Z
modelling cylindrical room or columns. The Cylinder2
statement, which creates a cylinder of the calotte type, will O

usually be preferable for modelling cylindrical ceilings. X

Y

The syntax for Cylinder is:

Cylinder <Number> <NumberOfSurfaces> <Radius> <RevAngle>

<Height> <T/B/N> <optional name>

<Number>
A unique number from 1 to 2,147,483,647 for identification of the first point and surface in the
Cylinder. Using the same number, but with negative sign defines the cylinder and its mirrored
counterpart in the XZ-plane (Y = 0). A Cylinder will take up several point- and surface numbers,
which must all be unique.

<NumberOfSurfaces>
For a full cylindrical room (with a revolution angle of 360°), around 16 to 24 surfaces are
recommended. For columns, a number between 6 to 8 is recommended.

<Radius>
Radius of the cylinder must always be greater than zero.

<RevAngle>
RevAngle must be within the range +/-360° and different from zero. If RevAngle is 180°, a half cylinder
is generated, and if it’s 360°, a full cylinder is generated. Positive revolution angles are defined
counter-clockwise.

<Height>
If the height is less than zero, the orientation of the cylinder is inverted. If height is zero, one circular
surface is generated.

Insertion point:
The insertion point of the cylinder is always the centre of the floor (bottom) surface.
Connection points:
The foot-points in Cylinder are stored in PListA
The top-points in Cylinder are stored in PListB

The example shown was generated with the following code:

CylinderStatement.Par
###
const N 16
const R 15
const H 10

Cylinder 1000 N R 270 H TB Cylindrical room

###

276
Hint! The cylinder can be made elliptical, using the MScale statement.

The Cylinder2 statement

Cylinder2 is a cylinder shell of the calotte type. Rather than specifying the radius and
revolution angle, Cylinder2 is specified in terms of the width and height. Cylinder2 is
typically used for cylindrical /curved ceilings.

The syntax for Cylinder2 is:

Cylinder2 <Number> <NumberOfSurfaces> <Width> <Height> <Length> <T/B/N>

<optional name> Z

O
Y
<Width> X

Width is oriented in the X direction on the Figure.

<Height>
Height of the cylinder shell is oriented in the Y direction on the Figure and may be positive (concave
shell) as well as negative (convex shell). Height must be different from zero and less or equal to
½*Width.

<Length>
Length of the cylinder shell is oriented in the Z direction the Figure. If Length is negative the
orientation is inverted.

Insertion point:
The insertion point of Cylinder2 is always foot point of the calotte floor (bottom) surface.

Connection points:
The foot-points in Cylinder2 are stored in PListA
The top-points in Cylinder2 are stored in PListB

The example shown was generated with the following code:

###
Const N 10
Const W 5
Const H 1
Const L 10

Cylinder2 1 N W H L TB Cylinder calotte

###

Hint! The cylinder can be made elliptical, using the MScale statement.

The Cone statement

The Cone statement models a cone. Typical use of the Cone
statement is for modelling half cone or cone shaped ceilings. Z

The syntax for Cone is: O

Y
Cone <Number> <NumberOfSurfaces> <Radius> <RevAngle> X
<Height> <optional name>

277
 <Number>
A unique number from 1 to 2,147,483,647 for identification of the first point and surface in the Cone.
Using the same number, but with negative sign defines the surface and its mirrored counterpart in
the XZ-plane (Y = 0). A Cone will take up several point- and surface numbers, which must all be
unique.

<Radius>
Radius of the Cone must always be greater than zero.

<RevAngle>
RevAngle must be within the range +/-360° and different from zero. If RevAngle is 180°, a half-cone is
generated, and if it’s 360°, a full cone is generated. Positive revolution angles are defined counter-
clockwise.

<Height>
The height must be different from zero. If the height is less than zero, the orientation of the cone is
inverted. Height is oriented in the Z-direction on the Figure.

The Cone example shown was generated with the following code:

###
const N 16
const R 15
const H 10

Cone 1 N R 270 H Cone shaped ceiling

###

Hint! The cone can be made elliptical, using the MScale statement.

The Dome statement

The Dome statement generates a full dome (half hemisphere)
covering the full 90° vertical angle. In most cases the Dome2
statement is probably better suited.
Z

O
The syntax for Dome is: X
Y

Dome <Number> <NumberOfSurfaces> <Radius> <RevAngle>

<optional name>

<Number>
A unique number from 1 to 2,147,483,647 for identification of the first point and surface in the Dome.
Using the same number, but with negative sign defines the dome and its mirrored counterpart in
the XZ-plane (Y = 0). A Dome will take up several point and surface numbers, which must all be
unique.

<Radius>
Radius of the Dome must always be greater than zero.

<RevAngle>
RevAngle must be within the range +/-360° and different from zero. If RevAngle is 180°, a half Dome is
generated, and if it’s 360°, a full Dome is generated. Positive revolution angles are defined counter-
clockwise.

278
 Connection points:
The right-side vertical points in Dome are stored in PListA
The left-side vertical points in Dome are stored in PListB
In the special case where the revolution angle is 180°, all points are stored in PlistA and the number
of vertical subdivisions is stored in ONVert.

The example shown was generated with the following code:

###
const N 16
const R 15

Dome 1 N R 270 This is a dome

###

Hint! The dome can be made elliptical, using the MScale statement.

The Dome2 statement

The Dome2 statement is a Dome shell of the calotte type, Z

where the vertical revolution angle is not necessarily 90°.

Rather than specifying the dome by a revolution angle, it is
O
specified by the width and height. Dome2 may typically be
Y
used for modelling dome shaped ceilings. X

The syntax for Dome2 is:

Dome2 <Number> <NumberOfSurfaces> <Width> <Height>

<RevAngle> <optional name>

<Number>
A unique number from 1 to 2,147,483,647 for identification of the first point and surface in the
dome. Using the same number, but with negative sign defines the surface and its mirrored
counterpart in the XZ-pane (Y = 0). A dome will take up several point- and surface numbers, which
must all be unique.

<NumberOfSurface>
Specifies the number of surfaces in one horizontal ring of the dome, around 16 to 24 surfaces per
ring is suggested. ODEON will automatically calculate the number of subdivisions in the vertical
level. If the revolution angle is 180° the number is stored in the ONVert variable, which would have
been 9 in the example above. The ONVert variable may help when connecting a Dome2 to a Cylinder2
in order to specify the correct number of surfaces in the cylinder.

<Width>
Width at the beginning of the dome. The width must always be greater than zero.

<Height>
The Height must be different from zero. If the height is less than zero, the orientation of the dome is
inverted. Height must be different from zero and less or equal to ½*Width.

<RevAngle>
RevAngle must be within the range +/-360° and different from zero. If RevAngle is 180°, a half-dome
is generated, and if it’s 360°, a full dome is generated. Positive revolution angles are defined
counter-clockwise.

279
 Connection points:
The right-side vertical points in Dome2 are stored in PlistA
The left-side vertical points in Dome2 are stored in PlistB

In the special case where the revolution angle is 180°, all points are stored in PlistA and the number
of vertical subdivisions is stored in ONVert.

The example shown was generated with the following code:

###
Const N 16
Const W 10
Const H 3
Const L 10

Dome2 1 N W H 270 Dome calotte

###

Hint! The cylinder can be made elliptical, using the MScale statement.

DebugIsOn and Debug

The debug options are useful when creating large or complicated geometries in the ODEON .par
format. Using these facilities can speed up geometry loading when loaded for preview only and
allow debugging of parameter values in geometry files. DebugIsOn is a Boolean, which can be set to
TRUE or FALSE, the syntax is:

DebugIsOn <Boolean>

Typically, you will insert the DebugIsOn flag in the beginning of the geometry file in order to
investigate parameter values, when loading geometry. When this Boolean is set to TRUE:

• ODEON will not prepare the geometry for calculation - as result the loading of rooms is
sped up.
• ODEON will enable debugging of parameters with the Debug statement.
The syntax for Debug is:

Debug <debug string>

In effect, anything can be put after the Debug keyword, i.e. you may put a complete copy of a line
in the .par file there. The contents following the Debug keyword is evaluated or if it can’t be
evaluated, it is echo'ed directly to the debug window in ODEON when loading the geometry and
it has no effect on the geometry. If DebugIsOn is set to FALSE, then debug lines are ignored. Contents
in the Debug strings, which cannot be evaluated, are displayed in quotes (").

Example: When loading the following .par file into ODEON:

###
DebugIsOn TRUE ;debug option turned on - if DebugIsOn is set to false then Debug lines are ignored
const L 6
Debug L ;debug a single constant
const W 4
Debug const W 4
const H 2.7
Debug L W H ; Debug values of L, W and H

Box 1 L W H TB Walls, floor and ceiling

Debug Box 1 L W H TB Walls, floor and ceiling ; Debug a complete line
###

280
ODEON will create this Debug window as a response:

Creating a new .Par file - time saving hints

The golden rule when creating a .Par file to model a room is to think carefully before you start
typing. For very simple rooms, it is not too difficult to keep track of things, but for realistically
complex rooms a systematic approach is desirable. You will typically have a set of drawings, which
have to be used as the basis for the ODEON model. It pays to spend quite a long time working out
how the room can be simplified to a manageable number of sensibly shaped plane surfaces,
sketching over the drawings. These ideas will have to be modified when you start to work out the
actual coordinates, to ensure that the surfaces really are plane. Here are some ideas that may help
you to create correct surface files faster:

• Exploit symmetry: If the room has an axis of symmetry, place the coordinate axis on it.
Then use the ‘sign'- convention for symmetric /semi symmetric modelling.
• If there are vertical walls and /or features, which repeat vertically (e.g. identical balconies),
use the CountPt, CountSurf, RevSurf statements or indeed For..End constructs.
• Build the room gradually, testing the .Par file at each stage of growth by loading it into
ODEON and have a look at the result.
• Use hybrid statements such as Box, Cylinder etc.

Where it is difficult to get surfaces to meet properly without either warping or using lots of small
surfaces to fill the gaps, allow the surfaces to cut through each other a little. This will usually ensure
a watertight result, which is more important, and has only minor drawbacks. These are (i) the
apparent surface area will be a little too big, affecting reverberation times estimated using Quick
Estimate Reverberation and the room volume estimated by Global Estimate, (ii) crossing surfaces
can look odd and hinder clarity in the 3D displays.

Do not try to include small geometrical details at the first attempt. If there are some large surfaces,
which are basically plane but contain complex geometrical features (e.g. a coffered ceiling), model
them at first as simple planes. Then once the room has been made watertight, make the necessary
alterations to the geometry file. The simplified version can also be used in the prediction exercises,
to give some idea of the effect of the feature in question.

Examples on parametric modeling

This section will give some short examples on the modelling of rooms using the parametric
modelling language of ODEON. The options in this modelling format are many, ranging from
typing the model number by number, to dedicated programming. This section will try to give an
idea on how to use the language and its keywords. In the default room directory created at the
installation of ODEON you may find several other examples in .Par format.

Four ways to model a box

These examples show four ways to model a box-shaped room; using plain numbers, using
constants, using constants plus symmetric modelling and using the Box statement along with the
MTranslate statement. In each example the dimensions of the room are: (W, L, H) = (4, 6, 2.7).

281
Below the box-shaped room is modelled using plain decimal numbers:

Parametric sample BoxFromPureNumbers.par

###
Pt 1 0 2 0
Pt 2 0 -2 0
Pt 3 6 -2 0
Pt 4 6 2 0
:ceiling points
Pt 11 0 2 2.7
Pt 12 0 -2 2.7
Pt 13 6 -2 2.7
Pt 14 6 2 2.7
Surf 1 floor
1 2 3 4

Surf 2 ceiling
11 12 13 14
Surf 3 end wall
1 2 12 11

Surf 4 end wall

1 2 12 11

Surf 5 side wall

1 4 14 11

Surf 6 side wall

2 3 13 12
###

Below the box-shaped room is modelled using constants for the definition of W, L and H. Some of
the advantages of using parameters in modelling rooms are that it makes it much easier to apply
changes to the model (allowing reuse) and often it will also improve the clarity of a model data.

Parametric sample BoxFromParameters.par

The box measures are:
Width = 4 metres
Length = 6 metres
Height = 2.7 metres
###
const W 4
const L 6
const H 2.7

Pt 1 0 W/2 0
Pt 2 0 -W/2 0
Pt 3 L -W/2 0
Pt 4 L W/2 0

Pt 11 0 W/2 H
Pt 12 0 -W/2 H
Pt 13 L -W/2 H
Pt 14 L W/2 H

Surf 1 floor
1>4

Surf 2 ceiling
11>14

Surf 3 end wall

1 2 12 11

Surf 4 end wall

282
 1 2 12 11

Surf 5 side wall

1 4 14 11

Surf 6 side wall

2 3 13 12
###

Below the box-shaped room is modelled using parameters and symmetric modelling syntax (signs
on point and surface numbers). The symmetric modelling syntax means less typing and less typing
errors:

Parametric sample BoxFromParametersUsingSymmetricModelling.par

###
const W 4
const L 6
const H 2.7

Pt 1 0 W/2 0
Pt 2 L W/2 0

Pt 11 0 W/2 H
Pt 12 L W/2 H

Surf 1 floor
1 2 -2 -1

Surf 2 ceiling
11 12 -12 -11

Surf 3 end wall

1 11 -11 -1

Surf 4 end wall

2 12 Mirror ;Mirror works just as well – defines point -12 and -2
Surf -5 side wall
1 2 12 11
###

Below the box-shaped room is modelled using the Box statement, which is the easiest way to create
this simple geometry. A MTranslate statement is used to insert the Box at the same position as in the
three other examples:

Parametric sample BoxStatement.Par

###
const L 6
const W 4
const H 2.7

MTranslate l/2 0 0

Box 1 l w h tb Walls and floor

###

Modeling a cylinder
This example shows two different ways to create a cylindrical room with a floor and a ceiling. In
the first example the room is modelled using the Cylinder statement:

Parametric Sample CylinderStatement.Par

###
const N 16
const R 15
const H 10

283
 Cylinder 1000 N R 360 H TB Cylindrical room
###

The Cylinder statement is of course the easiest way to model a cylinder, however sometimes more
flexibility is needed (e.g. different radius in top and bottom). In the second example the corners in
the room are modelled using the CountPt statement and the cylindrical surfaces are modelled using
the RevSurf statement. Notice that the number of points created by the CountPt statement is one
higher than the number of sections in the RevSurf statement. The bottom and top of the room is
modelled using the Surf statement, notice that points used by these surfaces are referenced using
the statement 100>100+N-1 rather than writing each of the sequential points. This is not only a faster
way to write things, it also allows a rapid change to the number of sections in the cylinder by simply
changing the N constant.

Parametric sample, a cylinder RevSurfCylinder.Par

###
const N 16
const R 15
const H 10

CountPt 100 N+1 R*CosD((PtCounter)*360/N) R*SinD((PtCounter)*360/N) 0

CountPt 200 N+1 R*CosD((PtCounter)*360/N) R*SinD((PtCounter)*360/N) H

RevSurf 300 100 200 Sections cylinder walls

Surf 100 Circular floor

100>100+N-1

Surf 2 Circular ceiling

200>200+N-1
###

Modelling a box-shaped room with columns in two dimensions, using two level For..End
constructs
When modelling geometries that have more than one level of symmetry, it is advantageous to use
For..End constructs. This example shows how to model columns in two dimensions in a room using
a two-level For..End construct.
Each column is created using 8 points and 4 surfaces, thus the numbering used by points and
surfaces is incremented by 8 each time a column is created. This is done by incrementing the
predefined variable NumbOffSet by eight for each column in order to make surface and point
numbers unique. The different positions of the points used for each column are obtained, using
MTranslate and MReset.

Parametric sample BoxColumnRoom.Par

###
const L 10
const W 4
const H 3
const NumColX 4
const NumColY 3
const ColumnW 0.3

MTranslate l/2 0 0
Box 1 l w h tb walls in the room
:modelling the columns

for ColYCnt 1 NumColY

MReset
MTranslate L/(NumColX+1) w/2-ColYCnt*W/(NumColY+1) 0
for ColXCnt 1 NumColX

284
 NumbOffSet NumbOffSet+8 ;comment: hint! setting NumbOffSet to Auto would do the same
job
Box 1 ColumnW ColumnW h n columns in the room
MTranslate L/(NumColX+1) 0 0
end
end
###

Using hybrid statements and coordinate manipulation

The following example demonstrates an example on how to use the hybrid statements Cylinder2
and Dome2 as well as the coordinate manipulations, which are essential for the use of the hybrid
statements.

This example is a rather complex one, so the main parts of the file is explained below:

Line 3-7 Defining constants.

Line 8-9 Inserting the cylindrical wall, which needs a rotation of 90° around the Z-axis.
Line 11 The foot-points of the cylindrical wall, which are temporarily stored in PListA, are
stored in PList0 for later use (definition of the floor).
Line 12-13 Inserting the dome shaped ceiling. The Z-rotation has already been set to 90°
when the wall was created.
Line 14-18 Setting the coordinate manipulation for the ceiling and creating the ceiling.
Line 19 Resetting the coordinate manipulation to work in absolute coordinates
Line 20-23 Creating wall/floor points
Line 24-25 Defining floor, using the ‘cylinder points’ stored in Plist0.
Line 28-29 Defining side walls using symmetric modelling.
Line 30-31 Defining back wall, using the ‘ceiling cylinder points’ which is still stored in
PListB.

1. Dome2 and cylinder2 x 2 room.par

2. ###
3. Const H 5
4. Const L 10
5. Const W 15
6. Const N 12
7. Const HCurve 4
Z

8. MRotateZ 90
9. Cylinder 1000 N W/2 180 H N O
10. //Stores PListA for later use with floor Y
11. PList0 PListA X

12. MTranslate 0 0 H
13. Dome2 2000 N W HCurve 180 Halfdome

14. MReset
15. MRotateZ 90
16. MRotateY 90
17. MTranslate 0 0 H
18. Cylinder2 3000 ONVert W HCurve L n Cylindric ceiling

19. MReset
20. Pt 1 0 W/2 0
21. Pt 2 L W/2 0
22. Pt 3 0 W/2 H
23. Pt 4 L W/2 H

24. Surf 1 Floor

25. 2 PList0 -2

26. //when done using PList0, it’s a good habit to Reset it

27. ResetPList0

285
 28. Surf -2 Side Walls
29. 1 2 4 3
30. Surf 3 BackWall
2 PListB -2
31. ###

Defining surfaces with concave edges

Most surfaces in the geometries used with ODEON will probably have convex edges (rectangles,
cylindrical surfaces etc.), however in ODEON, it is possible to define surfaces with cavities, even
surfaces with holes. Such surfaces are defined just like any other surface, by creating a list of corners
where the listing is obtained by travelling around the surface's edge (in either direction). Below are
two examples; one with a donut-shaped balcony floor and another with a cylindrical window
opening in a ceiling.
Z
In the donut example two ‘rings of corners’ are 207
208 206
created using the CountPt statement, notice that 108 107 106
209 205
the point 100 is equal to point 112 and point 200 109 105
is equal to point 212. The donut surface is 210 110 O 104 204
created, simply by connecting the inner and Y
111 X 103
outer ring of points into one surface. It doesn’t
211 203
matter whether one of the rings are created clock 100
112 102
101
or counter-clockwise. The surface is created from
200
212 202
the following list of points:
100,101,102,…,110,111,112,200,201,202,….,210,211,212. 201

DonutSurface.par
###
Const R1 10
Const R2 15
Const N 12
CountPt 100 N+1 R1*CosD(360*PtCounter/N) R1*SinD(360*PtCounter/N) 0
CountPt 200 N+1 R2*CosD(360*PtCounter/N) R2*SinD(360*PtCounter/N) 0

Surf 100 Donut surface

100>100+N 200>200+N
###

The window example shows how a cylindrical window opening is created in ceiling surface. The
interesting surface in this example is surface 1, the ceiling surface. The surface is created from the
following list of points:1,100,101,102,103….,111,112,1,2,3,4.

286
 CeilingWithWindowTube.par
###
Const R1 0.75
Const R2 0.5
Const N 12

Pt 1 1 1 0
Pt 2 1 -1 0
Pt 3 -1 -1 0 Z
Pt 4 -1 1 0
3
CountPt 100 N+1 R1*CosD(360*PtCounter/N)
R1*SinD(360*PtCounter/N) 0 108 107 106
CountPt 200 N+1 R2*CosD(360*PtCounter/N) 109 105 4
R2*SinD(360*PtCounter/N) 2
110 O 104
Surf 1 Ceiling
1 100>100+N 1>4 2 111 103
Y
100
112 102
RevSurf 2 100 200 N Window tube 101
X
Surf 100 Window /glass
200>200+N-1
1
###

287
Appendix H: Mathematical
Expressions
In this section, mathematical expressions used in the .Par modelling format and in the Room
acoustic parameter list are presented. Constants, variables, point numbers, surface numbers and
coordinates may be defined using mathematical expressions. Where integer numbers are expected
(Counter ranges in for…end loops, point, surface numbers, etc.), the results of mathematical
expressions are automatically rounded to the nearest whole number.

Operation Syntax Example

Addition + 2+5 = 7
Subtraction - 3-1 = 2
Multiplication * 2*3 = 6
Division / 4/2 = 2
Power Base^ Exponent 2^3 = 8
or or
Power(Exponent,Base) Power(3,2) = 8
Root Root(Index,Radicand) Root(3,8) = 2
Round Round(X) Round(2.67676) = 3
Truncation Trunc(X) Trunc(1.7) = 1
or
Int(X)

Sine of an angle in radians Sin(X) Sin(0) = 0

Cosine of an angle in radians Cos(radians) Cos(PI/4)= 0,707106781186547573
Tangent of an angle in radians Tan(radians) Tan(PI/4) = 1
Cotangent of an angle in radians Cotan(radians) Cotan(180) = 0
Hyperbolic Sine of an angle in radians Sinh(radians) Sinh(0) = 0
Hyperbolic Cosine of an angle in radians Cosh(radians) Cosh(0) = 1
Sine of an angle in degrees SinD(radians) SinD(90) = 1
Cosine of an angle in degrees CosD(degrees) CosD(0) = 1
Tangent of an angle in degrees TanD(degrees) TanD(45) = 1
Cotangent of an angle in degrees CotanD(degrees) CotanD(90) = 0
Inverse Sine in radians ArcSin(Y) ArcSin(-Sqrt(2)/2)*180/PI = -45
Inverse Cosine in radians ArcCos(X) ArcCos(Sqrt(2)/2)*180/PI = 45
Inverse Tangent in radians ArcTan(Y) ArcTan(1)*180/PI= 45
Inverse Tangent II in radians ArcTan2(X,Y) ArcTan2D(1,-1)*180/PI = - 45
Inverse Sine in degrees ArcSinD(Y) ArcSin(-Sqrt(2)/2)*180/PI = -45
Inverse Cosine in degrees ArcCosD(X) ArcCos(Sqrt(2)/2)*180/PI = 45
Inverse Tangent in degrees ArcTanD(Y) ArcTan(1) = 45
Inverse Tangent II in degrees ArcTan2D(X,Y) ArcTan2D(1,-1) = - 45
Exponential Exp(X) Exp(1)= 2.71828182845904509
Natural Logarithm Ln(X) Ln(2.718281828459045091) = 1
Logarithm base 10 Log10(X) Log10(100) = 2
Logarithm base 2 Log2(X) Log2(8) = 3
Square Sqr(X) Sqr(2) = 4
Square root Sqrt(X) Sqrt(2)= 1.41421356237309515
Radius Radius(A,B) Radius(3,4) = 5
Absolute value Abs(X) Abs(-2342) = 2342
Sign Sign(X) Sign(-2) = -1; Sign(0) = 0; Sign(3) = 1
Minimum of two numbers Min(X,Y) Min(23,12) = 12
Maximum of two numbers Max(X,Y) Max(23,22) = 23

288