Consultancy Report

Ref: 10903-R01

Submitted to: Prepared by:

Davies Roberts

Flare Audio LTD. Gergely Orosz

Unit 11, Chartwell Business Consulting engineer
Centre, 42 Chartwell Road, Lancing
West Sussex, BN15 8FB
United Kingdom
Tel: +44 (0)1903 761 000 Checked by:

Dr Mike Lower
Principal Consultant

Acoustic characterisation of Calmer technology prototypes

June 2020

ISVR Consulting, University of Southampton

Highfield Campus, Southampton SO17 1BJ, United Kingdom
Tel: +44 (0)23 8059 2162 Fax: +44 (0)23 8059 2728
Email: [email protected] Web: www.isvr.co.uk
Contents

1. Introduction .................................................................................................................... 1

2. Measurement set-up ....................................................................................................... 3

2.1 Test environment 3
2.2 Test equipment 3
2.3 Frequency response measurement with headphones 4
2.4 Total distortion measurement with headphones 5
2.5 Frequency response measurements with an external sound source 8

3. Measurement procedure ............................................................................................... 10

4. Results and data analysis ............................................................................................. 14

4.1 Frequency response measurements with headphones 14
4.2 Total Distortion measurements with headphones 18
4.3 Frequency response measurements with external source 22

5. Conclusions .................................................................................................................. 28

6. Appendix ...................................................................................................................... 30
6.1 Equipment list 30
6.2 Narrow band frequency response data 31
6.3 TD measurement results 34
6.4 A-Weighting values 36

7. References .................................................................................................................... 37

This report shall not be reproduced except in full, without the prior written consent of ISVR Consulting

ISVR Consulting Report 10903-R01 / June 2020 Contents

1. Introduction

ISVR Consulting was engaged by Flare Audio to perform acoustic tests on their new
Calmer technology prototypes. These measurements aimed to characterise the acoustic
response of this technology and to objectively assess its typical effect on the hearing
profile of the user of this device.

It is understood that Calmer technology is designed to manipulate the acoustic response

of the ear above 1,000 Hz, by altering the modal characteristics of the outer ear. In
particular, the technology aims to reduce naturally occurring resonances associated with
the geometry of the ear canal and pinna. In order to achieve this, these passive acoustic
devices contain an acoustically designed geometry with two opposing openings towards
the eardrums and external noise sources. The soft silicon material seals around the
entrance of the ear canal and the device partially covers the concha.

The devices can also be worn under circumaural and supra-aural headphones. In this
case, the technology aims to improve the headphone listening experience by controlling
the acoustic characteristics of the human ear coupled to the headphone.

During the time of this measurement project, the acoustic facilities of ISVR Consulting
were out of service, due to the COVID-19 lockdown period in the United Kingdom.
The test procedures were designed specifically to ensure that measurements do not rely
on free-field (anechoic) acoustic conditions.

The primary aim of the project therefore was to evaluate the acoustic effect of Calmer
devices under headphones using an Acoustic Head and Torso Simulator (HATS).
Frequency response functions (FRF) of 2 headphone specimens on the HATS were
measured with and without the Calmer devices fitted. FRF results were analysed
comparatively to demonstrate and quantify the acoustic effect of wearing the devices.

ISVR Consulting Report 10903-R01 / June 2020 Page 1

 Similar measurements were also taken with an external sound source to characterise the
acoustic effect of Calmer with a far field source of excitation. The acoustic effect of the
room was minimised using digital signal processing.

Any alteration in the transfer characteristics of the ear that Calmer achieves could also
influence the level of distortion perceived by the user of the technology. Distortion
measurements with and without Calmer were assessed using headphones. Whilst the
technology is not expected to control the amount of distortion generated by the
headphone itself, some distortion components falling in the modal frequency range
could be reduced at the ear drum, if the resonant behaviour of the system is attenuated
by the technology. In order to test the total perceived distortion with and without
Calmer, measurements were taken with a multi-tone stimulus signal.

The project aimed to quantify the effectiveness of the Calmer technology using
frequency response and distortion measurements. This report details the measurement
configuration and procedure, presents the results and provides an objective assessment
of the acoustic features of this technology. The report specifically does not discuss the
subjective experience of wearing the device and does not attempt to link any of these
with the test results. Detailed analysis and identification of the exact physical-acoustic
mechanisms of the device also fell outside the scope of this investigation.

ISVR Consulting Report 10903-R01 / June 2020 Page 2

2. Measurement set-up

All tests were performed by ISVR Consulting on the 29th and 30th April. This section
of the report details the equipment, test conditions and data processing in the project.

2.1 Test environment

All tests were carried out in a furnished, ordinary room with dimensions of
3.5 m x 3.0 m x 2.2 m. The side and top boundaries are plastered walls/ceiling, the floor
is covered by a thin layer of carpet. Besides a regular wooden door, the room is built
with double glazing (1.6 m x 1.1 m) on one of the walls. During all tests, windows and
the door were closed in order to minimise background noise levels. Furthermore, the
amplitudes of test signals were selected to ensure high signal-to-noise ratios. During
testing, external noise levels were observed and tests were halted if background levels
were higher than normal.

2.2 Test equipment

Tests were carried out using ISVR Consulting’s ‘Kemar’1 acoustic head and torso
simulator (HATS). This device is a manikin with a realistic head and pinnae and
incorporates ear simulators with ear canals and ‘eardrum’ microphones. It is
representative of a median human adult and is designed to permit acoustic
measurements of wearable acoustic devices. The shape of the manikin and the nature
of the ear simulators ensure that the obtained recordings include the acoustic effect of
the human body exposed to external sound sources or wearing headphones. For
headphones the manikin also creates a realistic acoustic load on the headphone drivers.
This platform is therefore ideal to investigate the changes that Calmer brings to this
system.

The ear simulators and microphones within the manikin are tuned to imitate the transfer
impedance of a typical human ear. ISVR Consulting’s ear simulators conform to
BS EN 60318-4:2010 2. The calibration of the HATS’s two microphones was checked

ISVR Consulting Report 10903-R01 / June 2020 Page 3

 before and after the measurement session using a Brüel & Kjær (B&K) type 4220
pistonphone. The calibrations were stable. The calibrations of the ear simulators,
microphones and pre-amplifiers are traceable to the manufacturer, G.R.A.S. who
verified their performance in March 2020. The B&K pistonphone was calibrated at a
UKAS accredited test house in March 2019.

The ear simulators accurately imitate the human ear’s (standardised) acoustic transfer
impedance up to the frequency of 10,000 Hz in compliance with BS EN 60318-4:2010.
Results above this frequency fall outside the range of the standard and may not be an
accurate human ear simulation, however comparative conclusions are made in this
report up to 16,000 Hz by relating recordings with and without Calmer in the exact
same conditions.

2.3 Frequency response measurement with headphones

The primary aim of this test was to measure and characterise how wearing Calmer
influences the sound pressure reaching the eardrum when used under headphones.
These measurements were performed by determining the impulse response between the
headphone excitation signal and the corresponding microphone recordings of the ear
simulator, with and without the Calmer prototypes ‘worn’ by the HATS under
headphones.

Measurements without Calmer provide the baseline or reference impulse response of

the headphones, against which any measurements with the device in place can be
compared. Comparisons of the impulse responses with and without the device indicate
the effect of these devices and show the changes in sound pressure levels (SPL)
obtained at the eardrum. The acoustic effect of wearing these devices under headphones
can be clearly determined using this method, since the only difference between
corresponding test conditions is the presence of the device.

ISVR Consulting Report 10903-R01 / June 2020 Page 4

 Brüel & Kjær’s Dirac 5.0 (Type 7841) was used in the FRF tests to measure the impulse
response of the complete electro-acoustic system from the electrical excitation signal
(for the headphone) to the ear simulator microphones with and without Calmer. In all
measurements, Dirac was set-up to produce a 10.9 second long exponential sine sweep
excitation signal without any source filter, sampled at 96 kHz.

To account for any imperfections imposed by the USB measurement interface used with
this software, Dirac’s sound device calibration routine was performed prior to the
measurements. This routine also ensured the input and output level calibrations are
obtained by the Dirac and the output levels therefore were automatically adjusted by
the software.

The exact calculation technique of this proprietary software is not detailed in its
reference manual, but it is understood that the Dirac software estimates the impulse
response by deconvolution of the ear simulator microphone signals and the
corresponding loudspeaker driving signal. This technique for estimating the impulse
response was established by A.Farina 3 at the 2000 AES convention in Paris.

The obtained impulse responses were processed using a 131072-point FFT analysis to
obtain the corresponding frequency response functions (FRF).

2.4 Total distortion measurement with headphones

The total signal distortion present at the ear drum microphones produced by the
headphones was measured with and without the Calmer devices using a Prism Audio
dScope M1 measurement system. A synchronous multi-tone measurement technique
was chosen specifically to perform distortion measurements in test conditions that can
emulate the complexity of real music and other programme material, whilst allowing
simultaneous distortion measurements in the whole audible range.

ISVR Consulting Report 10903-R01 / June 2020 Page 5

 Multi-tone testing was originally developed for rapid production line and broadcast
system testing; however, the technique holds several desirable features for this test
scenario. The reference manual of the measurement system used in the tests provides
limited information on the exact test parameters and calculation techniques, however
the system is understood to operate on the following principles:

- The test signal consists of 31 tones with frequencies close to, but not exactly
corresponding to 1/3rd octave centre frequencies as defined in BS EN 61260-
1:20145.
- The tones are identical in magnitude and their relative phases are varied randomly
to ensure that signal clipping is avoided, and that the crest factor of the signal is
controlled
- The frequencies of the tones are defined to avoid spectral overlaps of fundamentals
and harmonics. Furthermore, all frequency components of the input signal also
correspond to the exact centre frequency of an even FFT bin, allowing the use of
rectangular windowing without spectral leakage.
- The system in this configuration can detect the signal in the predefined FFT bins,
distortion in other even-numbered FFT bins and noise in all odd-numbered FFT
bins simultaneously

Exploiting these features of the multi-tone signals, this test can maximise the production
and detection of harmonic, and intermodulation distortion as well as other non-
linearities, which are more difficult to evaluate with more traditional swept-sine or
twin-tone tests. The complexity of the multi-tone test signal enables realistic test
conditions by better approximating the dynamic characteristics and tonal content of
music and other programme material on the whole audible range. Similar to real-life
listening, the headphones playing the multi-tones are expected to produce significantly
higher amounts of distortion products compared to those detectible in the FRF
measurements. Results are presented in this report using the following quantity:

ISVR Consulting Report 10903-R01 / June 2020 Page 6

 𝑠𝑠𝑠𝑠𝑠𝑠 𝑜𝑜𝑜𝑜 𝑎𝑎𝑎𝑎𝑎𝑎 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑟𝑟 †
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 (𝑇𝑇𝑇𝑇): 10 log10 � �
𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑒𝑒 𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 ∗

† calculated using spectral power magnitudes in even-numbered FFT bins that do not contain any components of the
input signal
∗ Reference used in TD calculation is the measured signal magnitude of the tone component closest to 1,000 Hz

N.B. TD results are not directly comparable with total harmonic distortion measurements

Since these tests aimed to establish the distortion characteristics of the signals measured
at the simulated eardrum rather than distortion generated by the headphones, sound
levels were not adjusted to give the equivalent undisturbed field sound levels. This test
relies on the assumption that the ear simulators used in this project provide a good
approximation to the behaviour of the human ear, both for the main (high magnitude)
signal components and the (low magnitude) distortion products. It is also assumed that
the instrumentation itself does not introduce a significant amount of its own distortion
that are typically not present in the human ear.

The reference power used in the distortion calculations may vary comparing cases with
and without Calmer if the corresponding frequency responses are changed by the
presence of the device at 1,000 Hz. Results were recorded with reference powers
calculated by the audio analyser, however individual correction factors were used in
calculations to ensure that the distortion readings are directly comparable.

Total Distortion results with and without Calmer were analysed comparatively. For
each test case, these metrics were logged for 30 seconds in the measurement system
and results in this report are calculated using linear averaging. The FFT results exported
from the analyser were also utilised for further calculations.

The Prism Audio multi-tone measurement system and calculation software complies
with requirements regarding multi-tone based distortion measurements set in BS EN
IEC 60268-21:2018 4.

ISVR Consulting Report 10903-R01 / June 2020 Page 7

2.5 Frequency response measurements with an external sound source

In order to characterise the acoustic effect of Calmer when worn without any other
acoustic devices around the ear, the impulse response measurements detailed in section
2.3 were performed with a studio-grade loudspeaker as an excitation source instead of
headphones. This test is designed to measure and characterise how wearing Calmer
influences the sound pressure reaching the eardrums in a certain sound field due to
external sources.

In order to ensure that such results are not affected by the acoustic response of a test
venue, these measurements would need to be performed under free-field acoustic
conditions. During the time of this project, it was not possible to utilise ISVR
Consulting’s large anechoic chamber. Measurements were carried out in the same room
as used for the headphone measurements and the approximate (simulated) free-field
conditions were achieved using digital signal processing.

The acoustic centre of the excitation source was placed 1.2 m away from the centre of
the HATS (midway between the two ears). The direction of the HATS was aligned so
both ears were the same distance from the centre of the loudspeaker.

The following steps were taken in order to minimise the influence of the room on the
measurement results:

1. Each measured impulse response was analysed individually to detect the ‘time-of-
arrival’ of the first distinct reflection from the boundaries of the room reaching the
simulated eardrums. Knowing the time-of-arrival of the direct sound from the
loudspeaker and that of the first reflection, a weighting window was constructed
using a sinusoidal function. The length of the window function was adjusted to be
shorter than twice the difference of the detected timings. The peak/centre of this

ISVR Consulting Report 10903-R01 / June 2020 Page 8

 window function was aligned with the arrival of the direct sound component from
the loudspeaker. The window function was applied to the impulse responses by
(element-wise) multiplication, removing major room reflections from the impulse
response.
2. Reference (impulse response) measurements of the excitation source were also
taken with an omni-directional microphone. For these tests, the HATS was removed
from the room and the reference microphone was placed 1.2 m from the loudspeaker
on its acoustic axis. The signal processing detailed in point 1 were performed on
these recordings as well.
3. The FRF magnitudes of the HATS and reference measurements were calculated
from the corresponding impulse responses using a 131072-point FFT.
4. The loudspeaker FRFs were used to allow correction for the characteristics (any
colouration) introduced by the loudspeaker itself and reducing the influence of any
room effect that windowing could not remove.

Using windowing to compensate for room reflections inherently compromises the

validity of results at lower frequencies. The length of the window used in this test
dictates the low frequency limit. For these results, any data below 450 Hz is invalid.

Measurements performed without Calmer are also known as Head-Related Transfer

Functions (HRTFs). In order to verify the effectiveness of the simulated free-field
processing, HRTFs obtained in these tests were compared to results of identical
measurements performed in the anechoic chamber in 2019. It was found that the real
and simulated free-field HRTFs matched between 450 Hz and 10,000 Hz with the
accuracy of ± 3 dB. This confirms the validity of the test set-up for this investigation.

Besides on-axis sound incidence described above, tests were also performed at 45o and
90o azimuth angles.

ISVR Consulting Report 10903-R01 / June 2020 Page 9

3. Measurement procedure

The aim of this test was to measure and characterise how wearing Calmer influences
the sound pressure reaching the eardrum. These measurements were performed by
determining the impulse response between the excitation source (loudspeaker or
headphone) and the corresponding microphone recordings of the ear simulator. Using
headphones, Total Distortion (TD) was measured with a multi-tone stimulus by
analysing the corresponding ear simulator signals on the frequency domain. In all
measured data pairs (with and without Calmer), every measurement conditions were
consistent, except the presence of Calmer in the artificial ears.

The tests were carried out using the following Headphone test specimen:
1. BOSE Soundlink AroundEar II Wireless
2. Nubwo N16 circumaural gaming headphone

Headphone FRF results were recorded using 20 mV rms signal level. Input levels in
TD tests were varied.

Loudspeaker measurements were carried out using a Genelec 8030C studio monitor.
Playback level was adjusted to be 75 dB at 1,000 Hz.

Any variability in the fitting of both the Calmer prototypes and/or the headphones on
the HATS can significantly influence the measurement results in all tests, particularly
at low and high frequencies. In order to minimise measurement errors related to the
fitting of the devices and the headphones, the following actions were taken:

1. The devices were examined and worn by the experimenter to determine the
properties of the typical fit the user of this device would achieve
2. Fit and seal of Calmer was visually examined and confirmed in the HATS’s ears
without the microphones in place

ISVR Consulting Report 10903-R01 / June 2020 Page 10

 3. Preliminary measurements were taken with approximately 10 refits to observe
variations that different fits introduce
4. Every refit of Calmer was visually examined to ensure proper seal and alignment
in the simulated ears
5. Every fit of the headphones was adjusted to be as symmetrical as possible on the
two ears, whilst ensuring that the pads fit on both ears with good seals all around.
Headbands of the headphones were adjusted to help achieve these fit criteria.

Once various fit properties were observed and reasonable consistency was achieved, 3
sets of FRF and 3 sets of TD measurements using headphones were performed with a
refit between every consecutive test. Using the left and right units individually,
presented results are based on the average of 6 (refitted) FRF measurements and 6
(refitted) TD measurements.

In the case of measurements with external sources, on-axis, 45o and 90o azimuth
(incidence of sound), measurements were performed consecutively. For each
positioning of the HATS, the reference HRTF was measured once (after preliminary
measurements to verify the setup), followed by 5 measurements with Calmer fitted in
the ears. Amongst these 5 measurements, the devices were refitted each time.

A typical fit of Calmer is shown on Figure 1.

ISVR Consulting Report 10903-R01 / June 2020 Page 11

 Figure 1: Typical fit of Calmer in HATS

A typical fit of the headphones are shown on Figure 2.

Figure 2: Typical fit of BOSE (left) and N16 (right) headphones on HATS

ISVR Consulting Report 10903-R01 / June 2020 Page 12

 It was observed in both cases, that minor variations in fit influenced the magnitude and
frequency of some high frequency resonant peaks. In order to allow averaging in these
conditions, the main FRF results are presented as 1/3rd octave smoothed data. All
presented averages are based on individual measurements, including averaging left and
right recordings in identical conditions (except off-axis measurements). This ensures
that any systematic errors introduced by minor positioning errors or other
inconsistencies are minimised. The processed dataset is normalised to 0 dB at 125 Hz
without the Calmer. The individual narrow band frequency responses with headphones
are shown in the Appendix.

ISVR Consulting Report 10903-R01 / June 2020 Page 13

4. Results and data analysis

4.1 Frequency response measurements with headphones

Figure 3 presents the magnitude of the measured frequency response function of the
BOSE headphones with and without Calmer technology applied.

Figure 3: FRF magnitude results of the BOSE headphones,

with and without Calmer

Figure 4 presents the magnitude of the measured frequency response function of the
N16 headphones with and without Calmer technology applied.

ISVR Consulting Report 10903-R01 / June 2020 Page 14

 Figure 4: FRF magnitude results of the N16 headphones,
with and without Calmer

Measurements presented on Figure 3 and Figure 4 are analogous to sound pressure

levels reaching the eardrum with and without Calmer from the headphones, assuming
that all frequencies are excited equally.

Figure 5 shows the effect of wearing Calmer normalised to the measurement result
without the devices. These results quantify Calmer’s influence on the sound pressure
levels reaching the eardrums for these two headphones.

ISVR Consulting Report 10903-R01 / June 2020 Page 15

 Figure 5: Insertion gain of Calmer devices with the two tested headphones

At frequencies below 1,000 Hz, Calmer does not significantly affect the measured
response. This may be explained by the fact that at these frequencies, the wavelengths
of the sound components are significantly longer than the dimensions of these devices.

Using the BOSE headphones with Calmer, no change in sound pressure levels were
observed up to 2,000 Hz. In case of the N16 headphones, Calmer had a minor boosting
effect on frequencies between 500 Hz and 2,000 Hz. The peak of 2.5 dB of this boosted
region was observed at 1,600 Hz.

Considering the measured results of the two headphones without Calmer, both
headphones show elevated responses in the 2,000 Hz to 8,000 Hz range. In this
frequency region the various dimensions of the headphones and outer ear become
ISVR Consulting Report 10903-R01 / June 2020 Page 16
 comparable with the wavelengths of the sound waves. Therefore, various wave
reflections from the headphones’ body, parts of pinnae and the eardrum lead to
modal/resonating acoustic characteristics in this frequency region.

Resonant mechanisms of the open/unaided ear are well established. Shaw and
Teranishi6 (1968) investigated the various components of head-related acoustic transfer
functions. Amongst various mechanisms related to the presence of head and shoulders,
for a free-field sound incidence of 45o azimuth, they identified resonances of
approximately 11 dB at 2,700 Hz and 8 dB at 5,800 Hz. Their research linked these
resonances to the geometry of the ear canal and concha respectively. When using
headphones and depending on the listeners exact ear geometry, the magnitude and
frequency of these resonances may change, however they will be present, influencing
the sound pressure reaching the eardrum. Results in this frequency region may also be
influenced by the frequency response characteristics of the headphone drivers
themselves.

According to test results on Figure 3 and Figure 4, Calmer attenuated these naturally
occurring acoustic phenomena and it was observed that this technology significantly
modified most of these effects. The responses of both headphones showed their highest
peak around 3,000 Hz (most likely due to simulated ear canal resonances). Calmer
attenuated these features, reducing levels by approximately 6.6 dB and 7.8 dB around
3,000 Hz respectively with the BOSE and N16 headphones.

The responses of both headphones showed further resonances in the 4,000 Hz to 8,000
Hz range. These are most likely linked to the geometry of the simulated concha and
other parts of the outer ear. Calmer achieved 6.4 dB and 6.7 dB of average attenuation
in this range with the BOSE and N16 headphones respectively.

Results in Figure 3 and 4 show that responses recorded with Calmer were lower than
those without this technology above 8,000 Hz. It is important to note that the response
of the ear simulators in the HATS are not defined above 10,000 Hz in BS EN 60318-

ISVR Consulting Report 10903-R01 / June 2020 Page 17

 4:2010, however comparative analyses of the 10,000 Hz, 12,500 Hz 16,000 Hz one-
third octave bands were performed, accepting that the two of these fell outside the usual
range. In these bands Calmer achieved an average attenuation of 4.4 dB (BOSE) and
3.9 dB (N16) compared to the corresponding cases without the devices.

4.2 Total Distortion measurements with headphones

Distortion measurements were performed using a multi-tone test method, as described

in section 2.4 of this report. Preliminary tests were carried out to assess the
characteristics of the distortion generated by the headphones and recorded by this test
method. The following observations were made with both headphones:

- Both headphones generated a significant number of distinct distortion products.

- FFT plots were analysed to identify 2nd to 7th harmonic distortion products of the
tones. This check identified a significant number of distortion products that are not
harmonically related to the input tones (confirming the effectiveness of the test over
swept sine measurements).
- Even bins, where distortion components were expected, were in most cases higher
than readings in neighbouring odd frequency bins. This indicates that most of the
detected distortion components were sufficiently above the noise floor of the
measurement.
- Highest distortion components were observed below 300 Hz, increasing towards
lower frequencies.
- Examining tendencies in distortion magnitude, on average, higher readings were
observed between 2,000 Hz and 8,000 Hz, compared to those between 300 Hz and
2,000 Hz.

The audio analyser used in the tests calculated total distortion during the tests as
detailed in section 2.4. Results for both headphones were recorded at 3 different

ISVR Consulting Report 10903-R01 / June 2020 Page 18

 input rms levels. In order to allow direct comparison of the distortion readings,
correction factors were used based on individually recorded levels at 1,000 Hz for
each measurement pair. In order to reflect the frequency dependent sensitivity of
the human hearing, total distortion and total signal power results were recorded with
an A-weighting filter applied to the inputs of the audio analyser.

Table 1 shows the average reduction in Total Distortion that Calmer achieved.
Individual distortion readings (with appropriate corrections applied) are shown in
Table A2 in the Appendix.
Table 1: A-weighted average reduction in Total Distortion (TD) by Calmer

BOSE AEII Wireless

Total reduction in
Total reduction in TD Total reduction in distortion by Calmer
Input Overall A-weighted
by Calmer A- signal power by A-weighted ref'd to
(mV SPL at eardrums
weighted (ΔTD) Calmer A-weighted reduction in signal
rms) without Calmer [dBA]
[dBA] (ΔSIG) [dBA] power (ΔTD/ΔSIG)
[dBA]

10 93.7 5.2 3.4 1.8

20 99.6 5.1 3.3 1.8
100 113.7 5.9 3.3 2.5
average N/A 5.4 3.3 2.1

N16 Gaming Headphone

Total reduction in
Total reduction in TD Total reduction in distortion by Calmer
Input Overall A-weighted
by Calmer A- signal power by A-weighted ref'd to
(mV SPL at eardrums
weighted (ΔTD) Calmer A-weighted reduction in signal
rms) without Calmer [dBA]
[dBA] (ΔSIG) [dBA] power (ΔTD/ΔSIG)
[dBA]
10 94.5 5.9 4.2 1.8
20 100.5 5.9 3.9 2.0
100 114.5 6.2 3.9 2.3
average N/A 6.0 4.0 2.0

Based on data corrected to common reference in dB calculation

ISVR Consulting Report 10903-R01 / June 2020 Page 19

 In all the tested cases, Calmer reduced the overall magnitude of the distortion
components reaching the eardrum. The highest reduction in total distortion of 6.2 dB
was observed with the N16 headphone at 100 mV rms input level.

The A-weighted total distortion results show an average reduction of 5.4 dB and 6.0 dB
with the BOSE and N16 headphones respectively. Reductions in corresponding signal
powers of 3.3 dB and 4.0 dB were recorded. In both cases, the reduction in distortion
power was greater than the reduction in signal power.

Identification and verification of the exact distortion reduction mechanism of Calmer

fell outside the scope of the project. However, based on the narrow-band FFT data used
in these calculations the following observations were made:

- Apart from distortion detected below 300 Hz, the frequency of the highest distortion
peaks without Calmer corresponded to the frequency of the resonant peaks observed
in Figure 3 and 4.
- Highest reductions in distortion components were found in the 2,000 Hz to 8,000
Hz range
- Analysis of distortion components in the 2,000 Hz to 8,000 Hz range, that fell close
to the corresponding signal tones showed comparable reduction to results observed
at the nearby tones
- Distortion components were present in a large number of even FFT bins across the
whole measured frequency range. A relatively small number of even FFT bins were
observed with levels comparable to odd frequency bins (containing noise only)

Using these observations, it is speculated that the primary acoustic mechanism behind
these results, is Calmer’s capability to attenuate the naturally occurring resonances in
the ear (as observed in the FRF results). In other words, Calmer reduces amplification
of distortion components at these resonant frequencies.

ISVR Consulting Report 10903-R01 / June 2020 Page 20

 Although the amount of distortion generated by the headphones without the acoustic
effect of the head and ears is not determined in this test, one can expect that the observed
resonances of the ear canal and pinna would boost the magnitude of distortion received
at the ear drums in the modal frequency range. This test demonstrated that a relatively
small number of signal components played by the headphones will excite a significantly
larger number of distortion components. It showed that Calmer’s effect of resonance
mitigation will affect both signal and distortion components. However, these tests
demonstrated that due to the significantly larger amount of distortion components the
overall reduction in distortion is greater than the overall reduction in signal components.

This test assumes that most real playback signals would behave similarly to the test
signal used, in terms of excitation and corresponding broadband distortion generation
mechanisms. However, the actual reduction in distortion relative to the signal
components experienced by the user will depend on the headphones used, the fit of the
devices and properties of the music or programme material reproduced.

It is important to note that this analysis is not exhaustive and is based purely on the data
collected in this project.

ISVR Consulting Report 10903-R01 / June 2020 Page 21

4.3 Frequency response measurements with external source

Figure 6 shows the magnitude of measured FRFs of Calmer technology applied to the
HATS with an external source place at 0o azimuth. Results without these devices are
also shown for comparison.

Figure 6: On-axis frequency response magnitude of Calmer measured using HATS

Figure 7 and Figure 8 show the magnitude of measured FRFs of Calmer technology
applied to the HATS with and external source place at 45o and 90o azimuth respectively.
The corresponding results without these devices are also shown for comparison.

ISVR Consulting Report 10903-R01 / June 2020 Page 22

 Figure 7: 45o azimuth sound incidence frequency response magnitude of Calmer
measured using HATS

ISVR Consulting Report 10903-R01 / June 2020 Page 23

 Figure 8: 90o azimuth sound incidence frequency response magnitude of Calmer
measured using HATS

These measurements are analogous to sound pressure levels reaching the eardrum with
and without Calmer from a point excitation source with a flat frequency response placed
in front of the listener (at 0°, 45° and 90° positions), in free-field conditions, assuming
that all frequencies are excited equally.

Figure 9 shows frequency response functions with Calmer normalised to the

corresponding FRFs without the devices. These results quantify Calmer’s influence on
the sound pressure levels (SPL) reaching the eardrums with external sound sources.

ISVR Consulting Report 10903-R01 / June 2020 Page 24

 Figure 9: Insertion gain of Calmer measured with an external source

Insertion gain results of Calmer with external sources of excitation presented in

Figure 9 show similar features to the results observed with headphones (near-field
sources of excitation) presented in Figure 5 of this report.

At frequencies below 1,000 Hz, Calmer does not significantly affect the measured
response. Frequencies between 500 Hz and 1,600 Hz are slightly boosted with the
devices. The highest amplification of 2.5 dB was observed at 1,250 Hz when the
loudspeaker was placed 90o off-axis relative to the HATS.

Resonant mechanisms of the human ear were briefly discussed in section 4.1 of this
report. HRTFs measured without the devices in Figures 6, 7 and 8 show the frequency
response of a typical, unaided human ear in the sound field established in this
experiment. It can be observed that the 3,000 Hz region is significantly enhanced by
ISVR Consulting Report 10903-R01 / June 2020 Page 25
 the presence of the human body and ear as well as that lower ranges show an increasing
trend towards 1,000 Hz. This observation is understood to be the combined effect of
numerous acoustic factors, of which the most significant are the natural resonances
occurring in the simulated ear canal, diffraction around the head, and reflections from
the pinnae and shoulders of the HATS, though shoulder reflections are minimised by
placing a tee-shirt on the HATS, as is recommended.

These results revealed that fitting Calmer to the HATS influenced these naturally
occurring acoustic phenomena. In the 2,000 Hz to 8,000 Hz range, the devices reduced
sound pressure levels measured at the eardrums. Comparative analysis of results with
and without Calmer were also performed in the 10,000 Hz, 12,500 Hz and 16,000 Hz
third octave bands. These high frequency bands also showed a reduction in sound
pressure levels compared to the corresponding HRTFs. Table 2 shows average
reductions observed at high frequencies.

Table 2: average reduction in SPL achieved by Calmer

SPL reduction with SPL reduction with

External source position Calmer, 2,000 Hz to Calmer 10,000 Hz to
8,000 Hz [dB] 16,000 Hz [dB]
00 azimuth 8.0 4.1
450 azimuth 7.6 5.2
900 azimuth 8.2 6.4

Results shown in Table 2 indicate comparable average attenuation figures at the 3 tested
source locations up to 8,000 Hz. It can be concluded that up to this frequency, the
acoustic effect of Calmer is reasonably uniform for ipsilateral sources in front of the
listener. At higher frequencies however, Calmer behaves more directionally with
moderately higher attenuation at wider source angles.

ISVR Consulting Report 10903-R01 / June 2020 Page 26

 Results presented in Figure 5 and Figure 9, along with the calculated average
attenuation data at high frequencies confirm the effectiveness of Calmer technology
with respect to Flare Audio’s design objective:

1. The devices do not influence the magnitude responses measured below 1,000 Hz
(below the resonant region of the outer ear) significantly.
2. The devices attenuate sound pressure levels reaching the ear drum above 1,000 Hz.
Peak SPLs due to natural resonant mechanisms observed in HRTFs in this
frequency region are significantly reduced.
3. The devices showed a comparable effect with near-field sources (headphones) and
far-field sources (external sounds and noises). Flare’s design target is met in both
cases. However, it was observed that the attenuation achieved with loudspeaker
sources were moderately higher compared to those measured with headphones.
4. With external sources, the devices did not significantly alter the directivity of
hearing (in the measured span of source locations)

ISVR Consulting Report 10903-R01 / June 2020 Page 27

5. Conclusions

Acoustic tests were performed on Flare Audio’s new Calmer technology prototypes.
Measurements were carried out using an acoustic head and torso simulator using two
different headphones and an external sound source. The acoustic effect of Calmer fitted
in the open ear and placed under the headphones was evaluated.

Frequency response measurements of the headphones and external source with and
without the technology prototypes revealed:

• No significant alterations in response were observed below 1,000 Hz when the

Calmer was fitted on the HATS
• Calmer reduces the magnitude of natural resonances observed in the HRTFs
(with external source) and in the headphone-to-eardrum acoustic system
between 2,000 Hz and 8,000 Hz.
• Between 2,000 Hz and 8,000 Hz Calmer achieved average attenuations of 7.9
dB and 5.9 dB with an external source and with headphones respectively
• Between 10,000 Hz and 16,000 Hz, the Calmer also decreased sound levels in
the ear with both source arrangements
• Results showed minor variations with the 2 different headphones used, however
results were comparable
• With external sources, the devices did not significantly alter the directivity of
hearing (in the measured span of source locations)
• The devices showed comparable effect with near-field sources (headphones)
and far-field sources (external sounds and noises).

ISVR Consulting Report 10903-R01 / June 2020 Page 28

 Total distortion measurements of the headphones with and without Calmer showed:

• Multi-tone stimulus tests showed an average reduction of 5.7 dB(A) in total

distortion power when Calmer was fitted under the headphones compared to
baseline recordings without the technology prototypes
• A-weighted measurements showed that the reduction in total distortion power
was greater than the reduction observed in the total A-weighted signal power of
the multi-tone stimulus
• It is speculated that the primary reduction mechanism behind these results is
Calmer’s capability to attenuate the naturally occurring resonances in the ear
(as observed in the FRF results)

The actual auditory experience and the effectiveness of wearing Calmer may vary
depending on the external sound field, type of headphone, fit and properties of the
programme material played.

ISVR Consulting Report 10903-R01 / June 2020 Page 29

6. Appendix

6.1 Equipment list

Table A1: Details of measurement equipment used

Equipment Manufacturer Type Serial number Measurement

HATS G.R.A.S KEMAR 1043 FRF, TD

Left ear coupler G.R.A.S RA0045 100378 FRF, TD
Left ear coupler G.R.A.S 40AG 88384 FRF, TD
microphone
Left ear G.R.A.S 26AC 86190 FRF, TD
microphone
preamplifier
Right ear G.R.A.S RA0045 100376 FRF, TD
coupler
Right ear G.R.A.S 40AG 88469 FRF, TD
coupler
microphone
Right ear G.R.A.S 26AC 86191 FRF, TD
microphone
preamplifier
Left pinna G.R.A.S KB0066 96746 FRF, TD
simulator
Right pinna G.R.A.S KB0065 96722 FRF, TD
simulator
HATS Brüel & Kjær Nexus 2572658 FRF, TD
microphone Type 2690
power supply
Loudspeaker Genelec 8030C 8030CP61122275 FRF
Reference Brüel & Kjær 4189 2539752 FRF
microphone
Measurement Creative X-Fi HD N/A FRF
frontend for Sound card
B&K software
Pistonphone Brüel & Kjær 4220 966195 FRF, TD
(Ear Coupler
calibrator)
Audio Analyser Prism Audio dScope 20040 TD
M1
Headphone FIIO A3 N/A FRF, TD
Amplifier

ISVR Consulting Report 10903-R01 / June 2020 Page 30

6.2 Narrow band frequency response data

Figure A1: Individual frequency response measurements of BOSE AEII headphone

ISVR Consulting Report 10903-R01 / June 2020 Page 31

 Figure A2: Individual frequency response measurements of N16 Gaming headphone

ISVR Consulting Report 10903-R01 / June 2020 Page 32

 Figure A2: On-axis frequency responses with and without Calmer with simulated
anechoic processing and loudspeaker response correction (data used for averaged
results)

ISVR Consulting Report 10903-R01 / June 2020 Page 33

6.3 TD measurement results

Table A2: TD measurement results of N16 headphone

N16
Calmer Input, mV TD Left TD Right
off 10 -26.1 -26.8
off 10 -25.5 -26.6
off 10 -26.5 -26.0
on 10 -31.8 -32.8
on 10 -31.7 -32.2
on 10 -30.7 -33.7

off 20 -21.4 -21.2

off 20 -19.8 -21.0
off 20 -18.1 -20.5
on 20 -24.8 -27.1
on 20 -25.0 -26.6
on 20 -26.6 -27.5

off 100 -8.5 -8.7

off 100 -7.2 -8.5
off 100 -5.6 -8.1
on 100 -13.0 -14.7
on 100 -13.4 -14.3
on 100 -12.7 -15.5

ISVR Consulting Report 10903-R01 / June 2020 Page 34

 Table A3: TD measurement results of BOSE headphone
BOSE
Calmer Input, mV TD Left TD Right
off 10 -34.1 -35.9
off 10 -33.4 -35.0
off 10 -33.7 -35.9
on 10 -39.2 -40.4
on 10 -39.2 -41.4
on 10 -38.9 -40.1

off 20 -28.5 -30.8

off 20 -28.2 -29.9
off 20 -28.4 -30.6
on 20 -34.1 -35.1
on 20 -33.6 -36.1
on 20 -33.5 -34.7

off 100 -16.6 -18.8

off 100 -16.3 -18.1
off 100 -16.7 -18.9
on 100 -23.7 -23.9
on 100 -22.4 -24.7
on 100 -22.0 -23.9

ISVR Consulting Report 10903-R01 / June 2020 Page 35

 6.4 A-Weighting values

1/1 Octave band centre frequencies 1/3 Octave band centre frequencies A-weighting values
Hz Hz dB

50 -30.2
63 63 -26.2
80 -22.5
100 -19.1
125 125 -16.1
160 -13.4
200 -10.9
250 250 -8.6
315 -6.6
400 -4.8
500 500 -3.2
630 -1.9
800 -0.8
1000 1000 0
1250 0.6
1600 1.0
2000 2000 1.2
2500 1.3
3150 1.2
4000 4000 1.0
5000 0.5
6300 -0.1
8000 8000 -1.1
10000 -2.5

ISVR Consulting Report 10903-R01 / June 2020 Page 36

7. References

1. Burkhard, M D, and Sachs, R M, (1975); Anthropometric manikin for acoustic

research. Journal of the Acoustical Society of America, 58, 214 – 222.
2. BS EN 60318-4:2010; Electroacoustics - Simulators of human head and ear. Part
4: Occluded-ear simulator for the measurement of earphones coupled to the ear by
ear inserts. British Standards Institution, London.
3. A.Farina, 2000; Simultaneous measurement of impulse response and distortion with
a swept-sine technique, Proceedings of the 108th AES Convention, February 2000
4. BS EN IEC 60268-21:2018; Sound system equipment. Acoustical (output-based)
measurements. British Standards Institution, London.
5. BS EN 61260-1:2014; Electroacoustics. Octave-band and fractional-octave-band
filters. Specifications. British Standards Institution, London.
6. Shaw EAG, Terenishi R. (1968): Sound pressure generated in an external-ear
replica and real human ears by a nearby point source. Journal of the Acoustical
Society of America 44:240-249

ISVR Consulting Report 10903-R01 / June 2020 Page 37