Professional Documents
Culture Documents
g
H
l
g
l
(2)
with g
l
C
M1
the free-eld Green functions for the M
microphones and the l-th scan point. Beamforming is
widespreadly used because it is easy to implement and
it is robust in presence of measurement noise. Typically,
for a unique and ideal point source in x
l
, i.e. cartesian
location of the l scan point, the source induces a CSM
given by
C = g
l
g
H
l
(3)
so that the lter-and-sum bemforming method gives
source powers Q
BF
ll
at scan point l
Q
BF
ll
= w
H
l
Cw
l
= w
H
l
[g
l
g
H
l
]w
l
. (4)
This is known as the Point Spread Function (PSF) and it
describes the microphone array and beamforming algo-
rithm response to a single ideal point source and it ex-
presses the emergence of side lobes. According to these
equations, we note that the predicted source power at
scan point l is exact Q
BF
ll
= 1 for l = l
. Ideally, we
would have Q
BF
ll
= 0 for l = l
Freq. [Hz]
P
o
w
e
r
/
F
r
e
q
.
[
d
B
r
e
f
1
/
H
z
]
)Welch Power Spectral Density Estimate, y = 20 dB
1
0
0
H
z
3
0
0
H
z
5
0
0
H
z
1
1
9
5
H
z
Target
Reproduced
(b) Microphone # 40
10
2
10
3
10
4
Freq. [Hz]
P
o
w
e
r
/
F
r
e
q
.
[
d
B
r
e
f
1
/
H
z
]
)
Welch Power Spectral Density Estimate, y = 20 dB
1
0
0
H
z
3
0
0
H
z
5
0
0
H
z
1
1
9
5
H
z
Target
Reproduced
Fig. 4: Power spectral density estimates of target and
reproduced sounds at two microphones. Test frequencies
for are shown as dashed vertical lines.
map at the studied frequency. This normalization al-
lows for the independent comparison of the sound in-
coming directions and source levels. From these g-
ures that provide left and right side views of the cylin-
drical scan grid, one clearly observes the shape corre-
spondence of the reproduced source power maps and the
target source maps. This evidence shows that our aircraft
cabin mock-up with sound eld rendering system is ef-
cient for recreating the spatial sound distribution of the
targeted sound environment. However, closer look and
comparison of the color scales highlight a source level
error that is sometimes positive or negative, for this case
it ranges from 4 to -3 dB. From other results [4] and from
the PSD shown in Fig 4 this was expected. In the case
of the 1195 Hz emerging tone, one can clearly note the
asymmetrical sound distribution of the target and repro-
duced source maps. This highlights the fact that the in-
strumented aircraft sound mock-up is not only able to
reproduce surrounding sound environment but localized
sources.
Figures 9 and 10 show similar source power maps ob-
tained for the same cases but with CLEAN-SC. At 100,
300 and 500 Hz, there is a clear correspondence between
the main source positions in the target and reproduced
acoustical source power map. Hence suggesting, as for
the focused beamforming results, that spatial sound re-
produced is efciently achieved. However, the target
map does not match well with the reproduced map at
1195 Hz. Although this would suggest a reduced spa-
tial accuracy, this CLEAN-SC result must be interpreted
with care. Indeed, the dirty maps (Fig. 5) from which
the CLEAN-SC map is obtained places the strongest
source on the left side and in front of the cylindrical scan
grid. However, the CLEAN-SC result at 1195 Hz for the
target sound eld placed the strongest source on the right
side and at the bottom of the cylinder. This suggests that
the iterative CLEAN-SC algorithm may have failed for
this test case. Accordingly, we suggest that the CLEAN-
SC algorithm may be more suited for localized-source
scenarios as it replace a maximum in the beamforming
dirty map by a localized source in a clean map. This
is further discussed in the next section.
6. DISCUSSION
From the reported experiments, on the basis of target
and reproduced source maps, we observed that the sound
eld reproduction inside the aircraft mock-up can pro-
vide an good spatial accuracy. It was also shown that
acoustic imaging is efcient for the objective evaluation
of spatial accuracy of reproduced sound elds and sound
environments using microphone array. One of the sim-
ple advantage is the possible independent comparison of
spatial sound distribution and power level of the spatial
distribution. Although this is a valuable tool for the com-
parison and evaluation of spatial source distribution, it is
not guaranteed that the acoustical source maps are the
real source distribution over the scan grid, which is, in
our case, the mock-up trim panels. Indeed, the acoustical
source power maps show the source distribution as seen
from the microphone array or heard from the listening re-
gion. A relevant result in this trend is that most of the re-
ported acoustical source power maps place the strongest
source on the front oor. This is also corresponds to
the scan points which are the farthest from the micro-
phone array and since we rely on normalized beamform-
ing using free-eld Green function (Eq. (2)) the geomet-
rical attenuation of the Green functions is compensated.
Therefore, for farthest scan points, the detected source
level might increased. For future experiment, we should
achieve similar tests but with a spherical scan grid around
the listening area, i.e. the microphone array, while as-
suming that listeners are immersed in a full surrounding
AES 52
ND
INTERNATIONAL CONFERENCE, Guildford, UK, 2013 September 24
Page 5 of 10
Gauthier et al. Evaluation of sound eld reproduction by acoustic imaging
sound environment. Furthermore, the use of a standard,
i.e. platform-independent, scan grid could then be more
easily transposed to other cases or vehicle types.
Besides the application of the tested algorithms and
microphone array to our specic applied case, namely
sound eld reproduction inside aircraft cabin mock-up,
the provided results allow for an extended discussion
about applicability of the method to spatial sound repro-
duction accuracy. It is rst and foremost adapted to the
evaluation of reproduced sound environment that have
been recorded using microphone array. Indeed, in this
case it is easy to apply similar acoustic imaging algo-
rithms to the recorded target array signals and recorded
reproduced array signals. Since array processing such
as classical beamforming introduces PSF that is heav-
ily dependent on the microphone array and scan grid ge-
ometries, it is more relevant to achieve such comparison
with the same microphone array, i.e. for the target sound
eld capture and recording of the reproduced sound eld.
Otherwise, then one could rely on CLEAN-SC algorithm
to ensure that the array-dependent PSF artifacts are re-
duced so that the spatial distortion of the specic array
is attenuated. However, in this paper we also shown that
CLEAN-SC might not be the most adapted method for
the objective evaluation of a surrounding sound environ-
ment with distributed sound source as CLEAN-SC tends
to replace distributed sound source by few single local-
ized sources.
The objective evaluation method tested in this paper,
could also be transposed to WFS that rely on the ac-
cumulation of simple theoretical sources for the virtual
creation of a full sound environment. This would need
some adjustments. Indeed, since WFS scenes involve
idealistic virtual sources, there is no target recording at
the microphone array that will be used for the recording
of the reproduced sound elds. Two approaches are pos-
sible: 1) theoretical simulation of the target sound eld at
the array using ideal virtual sources of the virtual scene
and 2) simple display of expected virtual source positions
over the reproduced acoustic image. For the latter case,
if proper spatial accuracy is reached, each main lobes of
the reproduced acoustic image should coincide with an
marked expected virtual position. For systems that are
aimed at the creation of precisely localized source po-
sition, as suggested previously, CLEAN-SC would be
more adapted and precise than classical beamforming.
For the case of WFS, there is an additional compromise
that is different from the paradigm of sound environment
reproduction in mock-up. Indeed, in this paper, we sim-
ply assumed that sound are emerging from the mock-
up inner boundary. This is a valid physical assumption
that dictated the shape of the scan grid. However, with
WFS, virtual source can be positioned at different dis-
tances hence introducing the issue of what the shape of
the scan grid should be. The scan grid should be de-
rived from a simple surface (sphere or cube) that simply
surround the listening area. For the evaluation of sys-
tem devoted to musical sound reproduction, we suggest
that spatial accuracy evaluation using acoustical imaging
should be done with broadband noise signals. Indeed,
this allows the use of the CSM matrix and averaging that
can provide a much more rened acoustic map.
7. CONCLUSION
Results of sound eld reproduction inside an aircraft
cabin mock-up were evaluated. The aim of the objec-
tive evaluation was the spatial accuracy of the reproduced
sound environment in comparison with the target sound
environment. The principal objective of this paper was to
evaluated the usability and efciency of acoustic imaging
using microphone array as a formal evaluation method of
spatial sound reproduction. Classical beamforming and
CLEAN-SC algorithms were compared. On the basis of
the provided results, we suggest that both algorithms can
be used. However, we suggest that CLEAN-SC is a bet-
ter choice for localized virtual source since it allows for
the distinction of tiny details in the acoustical image. For
distribution sound source reproduction, classical focus-
ing beamforming seems to be more appropriate. Besides
this general conclusion, the results presented in this pa-
per also shown that the instrumented cabin mock-up with
multichannel equalization is able to provide a reproduced
sound environment that is spatially very similar to the
target sound environment.
8. ACKNOWLEDGMENTS
This is a project involving: Consortium for Research and
Innovation in Aerospace in Qubec, Bombardier Aro-
nautique and CAE, supported by a Natural Sciences and
Engineering Research Council of Canada grant.
9. REFERENCES
[1] C.A. Powell, J.M. Fields, Human response to air-
craft noise, in Aeroacoustics of ight vehicles,
AES 52
ND
INTERNATIONAL CONFERENCE, Guildford, UK, 2013 September 24
Page 6 of 10
Gauthier et al. Evaluation of sound eld reproduction by acoustic imaging
Theory and practice, Volume 2: Noise control,
H.H. Hubbard Ed., Acoustical Society of America,
Woodbury, 1995.
[2] P.-A. Gauthier, C. Camier, F.-A. Lebel, Y. Pasco, A.
Berry, Experiments of sound eld reproduction in-
side aircraft cabin mock-up, presented at the 133rd
Audio Eng. Soc. Convention, San Francisco, 2012.
[3] C. Verron, P.-A. Gauthier, J. Langlois, C. Guas-
tavino, Spectral and spatial analysis/synthesis of
interior aircraft sounds, accepted for publication,
IEEE Transactions on Audio, Speech and Language
Processing (2013).
[4] P.-A. Gauthier, C. Camier, O. Gauthier, Y. Pasco, A.
Berry, Aircraft sound environment reproduction:
Sound eld reproduction inside a cabin mock-up
using microphone and actuator arrays, Presented
at the 21st International Congress on Acoustics,
Montral, 2013.
[5] D.A. McCurdy and R. E. Grandle, Aircraft noise
synthesis system, NASA technical memorandum
89040, 1987.
[6] K. Janssens, A. Vecchio, H. Van der Auweraer,
Synthesis and sound quality evaluation of exterior
and interior aircraft noise, Aerospace Science and
Technology 12 (2008) 114-124.
[7] N. Epain, E. Friot, G. Rabau, Indoor sonic boom
reproduction using ANC, Proceedings of Active
2004.
[8] S. Rizzi and B. Sullivan, Synthesis of virtual envi-
ronments for aircraft community noise impact stud-
ies, in 11th AIAA/CEAS Aeroacoustics Conference
(2005) 2291-2306.
[9] H. Ploner-Bernard, A. Sontacchi, G. Lichteneg-
ger, S. Vssner, Sound-system design for a pro-
fessional full-ight simulator, presented at the 8th
International Conference on Digital Audio Effects
(DAFx05), 2005.
[10] M. Keller, A. Roure, F. Marrot, Acoustic eld re-
production for psychoacoustics experiments: appli-
cation to aircraft interior noise, Proceedings of Ac-
tive (2006).
[11] E. Bongini, Modle acoustique global et synthse
sonore du bruit dun vhicules : application aux
vhicules ferroviaires, PhD Thesis, Universit de
Provence, France (2008).
[12] P. Basilio, J.J. Lpez, J. Escolano, L. Hrchens,
Multiactuator panels for wave eld synthesis:
Evolution and present developments, J. Audio
Eng. Soc. 58 (2010) 10451063.
[13] E.N.G. Verheijen, Sound reproduction by wave
eld synthesis, Ph.D. thesis, Technische Univer-
siteit Delft, 1998.
[14] M.J. Strau, D. de Vries, Application of multi-
channel impulse response measurement to automo-
tive audio, presented at the 125th Audio Eng. Soc.
Convention, Sans Francisco, 2008.
[15] E.G. Williams, Fourier Acoustics Sound Radi-
ation and Neareld Acoustical Holography, Aca-
demic Press, San Diego, 1999.
[16] J. Nowak, M. Strau, Sound eld reproduction
analysis in a car based on microphone array mea-
surements, presented at the 48th AES International
Conference: Automotive Audio, Munich, 2012.
[17] H. Kutschbach, Verication for spatial sound sys-
tems, presented at the AES 24th International Con-
ference, Canada, 2003.
[18] H. Teutsch, Modal Array Signal Processing: Prin-
ciples and Application of Waveeld Decomposition,
Springer, Berlin, 2007.
[19] S. Krober, K. Ehrenfried, L. Koop, A. Lauterbach,
A. Henning, Systematic comparison of micro-
phone array measurements in open and closed wind
tunnels, 16th AIAA/CEAS Aeroacoustics Confer-
ence, AIAA-2010-3734, Sweden, 2010.
[20] T. Padois, C. Prax, V. Valeau, Numerical vali-
dation of shear ow corrections for beamforming
acoustic source localisation in open wind-tunnels,
Applied Acoustics, 74 (2013) 591-601.
[21] P. Sijtsma, CLEAN based on spatial source coher-
ence, Aeroacoustics 6 (2007) 357374.
[22] P.-A. Gauthier, C. Camier, O. Gauthier, Y. Pasco,
A. Berry, Sound eld reproduction of real ight
recordings in cabin mock-up, presented at the AES
52nd International Conference, UK, 2013.
AES 52
ND
INTERNATIONAL CONFERENCE, Guildford, UK, 2013 September 24
Page 7 of 10
Gauthier et al. Evaluation of sound eld reproduction by acoustic imaging
Fig. 5: Acoustical source power maps [dB] correspond-
ing to the target sound eld obtained using focused
beamforming at 100, 300, 500 and 1195 Hz (left side).
Fig. 6: Acoustical source power maps [dB] correspond-
ing to the reproduced sound eld obtained using focused
beamforming at 100, 300, 500 and 1195 Hz (left side).
AES 52
ND
INTERNATIONAL CONFERENCE, Guildford, UK, 2013 September 24
Page 8 of 10
Gauthier et al. Evaluation of sound eld reproduction by acoustic imaging
Fig. 7: Acoustical source power maps [dB] correspond-
ing to the target sound eld obtained using focused
beamforming at 100, 300, 500 and 1195 Hz (right side).
Fig. 8: Acoustical source power maps [dB] correspond-
ing to the reproduced sound eld obtained using focused
beamforming at 100, 300, 500 and 1195 Hz (right side).
AES 52
ND
INTERNATIONAL CONFERENCE, Guildford, UK, 2013 September 24
Page 9 of 10
Gauthier et al. Evaluation of sound eld reproduction by acoustic imaging
Fig. 9: Acoustical source power maps [dB] correspond-
ing to the target sound eld obtained using CLEAN-SC
at 100, 300, 500 and 1195 Hz (left side).
Fig. 10: Acoustical source power maps [dB] corre-
sponding to the reproduced sound eld obtained using
CLEAN-SC at 100, 300, 500 and 1195 Hz (left side).
AES 52
ND
INTERNATIONAL CONFERENCE, Guildford, UK, 2013 September 24
Page 10 of 10