Practical Ambisonics

Practical Ambisonics
Enrique Tomás & Manu Mitterhuber

This presentation was created in May 2023 for the workshop developed at Musrara School of Arts and Society in Jerusalem.
It contains screenshots of IEM Ambisonic Plugins and images from the open access book Ambisonics: A Practical 3D Audio Theory for Recording,
Studio Production, Sound Reinforcement, and Virtual Reality by Franz Zotter and Matthias Frank published at Springer. These images are
reproduced for educational use only.
Spatial
Audio
Can
we capture it?
Can we
reproduce it?
Sound sources have spatial
information, how can we capture it?
Dealing with Spatial
Audio: Approaches
➔ Channel Based
A channel-based mix, usually for
a specific target loudspeaker
➔ Object Based
Encoding each audio source
independently with positional
metadata, and letting the
renderer at the reproduction side
position the audio as best it can
to the desired location.
Modern Spatial
Audio Renderers
(Panners)
➔ VBAP (Vector based amplitude panning)
Interaural level difference (ILD)
based on triangles. Only three
speakers are used at maximum.
➔ Ambisonic Panning
The audio scene is modelled as a
sphere (non-preferred direction).
There is always sound played in
all speakers.
Find the differences between
these renderers
Energy Spread Playback Systems
➔ VBAP maximizes sound ➔ VBAP renderers need to know
accuracy at the expense of all speakers positions. They
variability of the energy output the actual audio to be
spread: it is maximal at the played at each of the speakers.
middle of a loudspeaker ➔ The ambisonics signal is
triangle. decoupled from the playback
system. Audio is encoded in an
➔ Ambisonics has a constant intermediate format which is
energy spread regardless of independent from speaker
the source direction. The layouts. Playback requires a
decoder to transform this
lower the ambisonic order, ambisonic format to the final
the greater the spread. audio for each speaker.
Why encoding? For
capturing a spatial
soundfield in a way
which is independent
from playback layouts
We can compose
music independently
from the speaker
layout where it will be
played.
Which layouts?
Ambisonics benefit from spherical and
isomorphic layouts, but not necessarily…
Ambisonics Workflow
Audio contents are first encoded in ambisonic format and
later decoded to a particular speaker layout. Therefore, we
produce ambisonic scenes which can be played at different
setups.
The ambisonics encoded audio scene is the decomposition of an audio signal into a number
of spherical directions. For example, here a mono track is encoded at a position in space
(see blue sphere) resulting into a 25 channels audio signal. That means the mono file is
encoded using 24 spherical directions plus an omni channel.
The transformer can be any ambisonics effect or processing: reverbs, EQ, rotations,
spread effects, room effects, etc which are applied to the ambisonics signal (not to the
signal we will decode later).
The number of ambisonics audio channels usually does NOT coincide with the
number of speakers used (output channels). It depends on the ambisonic order. Here
we add a decoder for rendering the output audio channels to be played at the actual
speaker layout.
The number of ambisonics audio channels usually does NOT coincide with the
number of speakers used (output channels). It depends on the ambisonic order. Here
we add a decoder for rendering the output audio channels to be played at the actual
speaker layout.
Spatial Order
We can define the number

of spatial directions in the
ambisonics mix. The
smaller the order, the more
“pixelated” our spatial
resolution will be.
Ambisonics sounds good if
order is 3 or bigger.
Spatial Order
We can define the number of

spatial directions in the ambisonics
mix. The smaller the order, the
more “pixelated” our spatial
resolution will be. Ambisonics
sounds good if order is 3 or bigger.
Spatial Orders Higher Order:
decomposition into more spatial directions (harmonics)

To remember
➔ Spatial harmonics
The more spatial harmonics (the higher
the order), the less “spatially pixelated”
your soundfield will be.
➔ Which order?
We would love to use 1000th order but
we do not count with pragmatic software
or hardware. From 4th order or more
(HOA – High Order Ambisonics), the
quality improves substantially.
➔ Number of channels pro Order
Formula: (order + 1)2
i.e. 4th order = 25 channels
Spatial Orders Extending the M-S format
Spatial Orders Relationships between order and channels
Result: 4 channels ->three spherical axis X, Y, Z plus an omnidirectional

component W ->> First Order Ambisonics
Example First Order: decomposition in three spatial
Spatial Orders directions
Result: 4 channels ->three spherical axis X, Y, Z plus an omnidirectional

component W
Let’s practice
ambisonics!
Open Reaper
and the example
project
Workflow Encoder-Decoder
● Add “source” tracks as children of the
Ambisonics Bus (mono /stereo/
multichannel)
● Each “source” track must have its
MultiEncoder VST in FX.
● The “Ambisonics Bus” track receives and
mixes all source tracks which are already
ambisonic after their MultiEncoders.
● At the “Ambisonics Bus” you can add
ambisonics effects (reverb, compression,
EQ, etc).
● The “loudspeakers” track receives audio
channels from the “Ambisonics Bus” and
decodes them to the actual speaker rig.
● The “Binaural Decoder” track receives
from the “Ambisonics Bus” and produces
binaural decoded version of the mix.
Using the Encoder
● Select number of sound
sources
● Select SN3D format and the
ambisonics order
Defining a decoder
● Select order
● Introduce your speaker
positions
● Introduce “imaginary”
speakers for helping the
decoder algorithm
● Export to save your
decoder
● Import to load new ones
Using the
decoder
● We create decoders
with “AllRad” but we
prefer using the
“SingleDecoder” (after
loading a decoder)
because it incorporates
subwoofer output and
frequency band control.
Using several types of
sources: mono, stereo,
multichannel, B-
Format
● Practically, you will need to mix
different types of sound sources,
their encoded versions, and other
ambisonic native audio files.
Building a Scene
Case study: The Murder of Crows
Automation in Reaper
● Control multiEncoder and record each source’s position with
automation
Automation in Reaper: Modes
● Use the mode Latch to record a take of your
automation.
● Change to read for protecting (not deleting) the
previous take
● Shape the take drawing on the envelopes
Record MultiEncoder:
● with the mouse
● with Parameter Modulation
● with USB controller
● with OSC connections
Ambisonic Effects
EQ
Energy Visualiser
Compressor
FDN Reverb
Room Encoder
Etc
Exporting your
project
Render the
ambisonics bus
with a number
of channels
Alternatives in SPARTA
plugins suite
IR impulse (matrix convolver)
Encoder/Decoder
Upmixer
Practice!
Work Analysis!

Practical Ambisonics

Uploaded by

Copyright:

Available Formats

Practical Ambisonics

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Practical Ambisonics

Uploaded by

Copyright:

Available Formats

Practical Ambisonics

Enrique Tomás & Manu Mitterhuber

We can define the number

We can define the number of

decomposition into more spatial directions (harmonics)

Result: 4 channels ->three spherical axis X, Y, Z plus an omnidirectional

Result: 4 channels ->three spherical axis X, Y, Z plus an omnidirectional

You might also like