Modeling Auditory Localization of Subwoofer Signals in Multi-Channel Loudspeaker Arrays

1
An automated calibration system for telematic
music applications

Jonas Braasch Communication Acoustics and Aural
Architecture Research Laboratory
(CA3RL) Rensselaer Polytechnic Institute, Troy,
NY http//symphony.arch.rpi.edu/carl
2
Steinberg and Snow (1934)
Location A
Location B
3
(No Transcript)
4
Biggest Challenges

• Bandwidth
• Transmission latency
• Feedbacks
• Communication during setup.
• you will need to make friends with the sysadmin

5
Necessary Bandwidth

• Telematic Circle
• The transmission of DV quality video requires a
  bandwidth of 25 Megabits/s
• 8 channels of CD quality audio about 5.5
  Megabits/s.
• McGill University (Jeremy Cooperstock)
• Gigabit/s AV connection for HD uncompressed
• Fraunhofer-Erlangen
• New Fraunhofer/Erlangen low-latency coder AAC-LD
  (about 5 ms for the coding/decoding process),
  will be integrated into CISCOs system.

6
Transmission Delay

• Acceptable latency for music 25 milliseconds.
• Speed of light
• Signal traveling between RPI in Troy, NY and
  CCRMA at Stanford University, Palo Aalto, CA
  14 ms for the distance of 4,111 km (direct line).
  
• 54 ms for a connection between New York and
  Australia (16,000 km)
• Speed of sound
• 14 ms6 m (c430 m/s at room temperature)
• Total delay
• transmission delay (determined by physical
  distance propagation speed of the signal)
• signal-processing delay

7
Feedback loop in PA systems
8
Feedback loop in telematic connections
Longer transmission delays are easier to detect
audibly!
9
Echo feedback

• Audible colorations and echoes are a common side
  effect in two-way transmission systems.
• Audio/videoconferencing systems such as iChat or
  Skype use echo-cancellation systems to suppress
  feedbacks.
• In speech communication echo-cancellation systems
  work well, since the back-and-forth nature of
  spoken dialogue usually allows to temporarily
  suppress the transmission channel in one
  direction.
• In simultaneous music communication, however,
  this procedure tends to cut-off part of the
  performance.
• Solution capture music signals with closely
  spaced microphones (e.g., lavalier microphones).

10
Telematic Circle Members

• Deep Listening Institute
• Pauline Oliveros, Sarah Weaver
• Rensselaer Polytechnic Institute
• Curtis Bahn, Jonas Braasch, Pauline Oliveros
• Stanford University (CCRMA)
• Chris Chafe, Ge Wang
• University of California San Diego
• Mark Dresser, Shahrokh Yadegari, Adriene Jenik,
  Victoria Petrovich
• McGill University
• Jeremy Cooperstock, Bill Martens

11
Telematic Circle Concerts

• March 22 RPI, Northwestern, Stanford
• June, 26 ICAD 2007, Montreal
• McGill, Korea, RPI, Stanford
• Aug 6-7, SIGGRAPH 2007, San Diego
• UCSD, RPI, Stanford/Banff
• Nov 16, 2007 Stanford, RPI, UCSD
• Dec 14, ISIM 2007 RPI, Northwestern
• Aug 28, ICMC 2008, RPI, Belfast, Stanford

12
Telematic transmission scheme
13
from EMPAC webpage
14
Problem

• Automatic tracking of actors required during a
  theater film shoot
• Typically, the actors are recorded closely with
  lavalier microphones
• good dry sound quality
• ? Spatial aspects of the recordings are lost and
  need to be recovered

15
Possible Solutions to track actors

• Optical tracking works well but can fail in
  bright stage light
• GPS only works outdoors
• Electromagnetic tracking (e.g., via sender for
  lavalier mic) works only outdoors (reflections
  occur indoors)
• Acoustic tracking works well for single sound
  sources, problematic in multiple sound source
  scenarios

16
Acoustic Tracking Solutions

• Beamforming
• Methods based on interchannel delay
• Both methods work only well for single sound
  sources
• Need to find time-frequency windows in which only
  one sound source is present.
• Lavalier microphone data can be used for this
  purpose.

17
Sketch of the recording and reproduction set-up
ViMiCVirtual Microphone Control
(Multichannel sound spatialization software)
18
Microphone Array

• Microphone array was positioned in the center of
  the room, 186 cm above the ground.
• consists of 5 omni-directional microphones
  (Earthworks M30).
• square-based pyramid dimensions base side 14
  cm, triangular side 14 cm.

19
SNR Estimation
20
Estimation of the signal-to-noise ratios for each
sound source
21
Test material

• Material recorded during a theatre production.
• Room acoustics was typical for a small theatre
  and not reverberation free.
• Four actors were equipped with lavalier
  microphones.

22
Analyses methods

• programmed in MATLAB
• A filter bank of five octave-band wide IIR
  filters (Chebyshev Type-I filters, 125 Hz to 2
  kHz center frequencies.
• A running time-window was applied (Hanning,
  100-ms filter length). For each time/frequency
  bin, the signal-to-noise ratio was determined
  according to Eq. 1.
• Sound source estimation (if SNR gt 4 dB)
• cross-correlation technique
• information theoretic delay criterion (ITDC)
  algorithm Mod88

23
DOA for the left/right angle
with the speed of sound c, the sampling frequency
fs, the internal delay ?, and the distance
between both microphones d
DOADirection of Arrival
24
(No Transcript)
25
Results for cross-correlation technique
26
Results for cross-correlation technique
27
(No Transcript)
28
Results for ITDC
29
Results for ITDC
30
(No Transcript)
31
Conclusions

• Initial results promising.
• Significant improvement was observed when the
  cross correlation method was replaced with the
  information theoretic delay criterion algorithm.
• A real-time application will be implemented in
  future

32
Literature

• Fri80 Fritsch, F. N. and R. E. Carlson,
  Monotone Piecewise Cubic Interpolation, SIAM J.
  Numerical Analysis, Vol. 17, 1980, pp.238246
• Jot92 Jot, J.-M. (1992) Étude et réalisation
  d'un spatialisateur de sons par modèls physiques
  et perceptifs, Doctoral dissertation, Télécom
  Paris.
• Mod88 R. Moddemeijer, An information
  theoretical delay estimator, Ninth Symp. on
  Information Theory in the Benelux, May 2627,
  1988, Mierlo (NL), pp. 121128, Ed. K.A.
  Schouwhamer Immink, Werkgemeenschap Informatie-
  en Communicatietheorie , Enschede (NL).
• Pul97 Pulkki, V. (1997) Virtual sound source
  positioning using vector base amplitude panning,
  J. Audio Eng. Soc. 45, 456466.
• Wür97 W. Würfel (1997) Passive akustische
  Lokalisation passive acoustical localization,
  Master's Thesis, Technical University Graz.

33
Outlook

• Expand telematic Circle (contact braasj_at_rpi.edu)
• Integration of haptics into the transmission
  system
• Get Acoustic Tracking system to work in real-time

34
Room Model
35
AVE Architecture
36
(No Transcript)
37
AV integration
Valente Braasch, Acustica 2008
38
virtual sound source
virtual microphones
floor
39
Blumlein XY-Technique
Signal left channel
Signal right channel
Angle of arrival
Directivity pattern right microphone
Directivity pattern left microphone
40
Blumlein XY-Technique
Signal left channel
Signal right channel
45
Angle of arrival
Directivity pattern right microphone
Directivity pattern left microphone
41
Blumlein XY-Technique
Signal left channel
Signal right channel
45
45
Angle of arrival
Directivity pattern right microphone
Directivity pattern left microphone
42
Blumlein XY-Technique
Signal left channel
Signal right channel
45
0
45
Angle of arrival
Directivity pattern right microphone
Directivity pattern left microphone
43
Stereophony
gtgt
l-30
r30
a
a