Reduced-bandwidth and distributed MWF-based noise reduction algorithms Simon Doclo, Tim Van den Bogaert, Jan Wouters, Marc Moonen Dept. of Electrical Engineering.

Slides:



Advertisements
Similar presentations
Advanced Speech Enhancement in Noisy Environments
Advertisements

Acoustic design by simulated annealing algorithm
Adaptive Filters S.B.Rabet In the Name of GOD Class Presentation For The Course : Custom Implementation of DSP Systems University of Tehran 2010 Pages.
Microphone Array Post-filter based on Spatially- Correlated Noise Measurements for Distant Speech Recognition Kenichi Kumatani, Disney Research, Pittsburgh.
Speech Enhancement Based on a Combination of Spectral Subtraction and MMSE Log-STSA Estimator in Wavelet Domain LATSI laboratory, Department of Electronic,
3/24/2006Lecture notes for Speech Communications Multi-channel speech enhancement Chunjian Li DICOM, Aalborg University.
An Overview of Delay-and-sum Beamforming
Zhengyou Zhang, Qin Cai, Jay Stokes
Hearing & Deafness (3) Auditory Localisation
Spectral centroid 6 harmonics: f0 = 100Hz E.g. 1: Amplitudes: 6; 5.75; 4; 3.2; 2; 1 [(100*6)+(200*5.75)+(300*4)+(400*3.2)+(500*2 )+(600*1)] / = 265.6Hz.
1 New Technique for Improving Speech Intelligibility for the Hearing Impaired Miriam Furst-Yust School of Electrical Engineering Tel Aviv University.
Acoustical Society of America, Chicago 7 June 2001 Effect of Reverberation on Spatial Unmasking for Nearby Speech Sources Barbara Shinn-Cunningham, Lisa.
1 Recent development in hearing aid technology Lena L N Wong Division of Speech & Hearing Sciences University of Hong Kong.
Normalised Least Mean-Square Adaptive Filtering
Dept. E.E./ESAT-STADIUS, KU Leuven homes.esat.kuleuven.be/~moonen/
Speech & Audio Processing - Part–II Digital Audio Signal Processing Marc Moonen Dept. E.E./ESAT-STADIUS, KU Leuven homes.esat.kuleuven.be/~moonen/
A VOICE ACTIVITY DETECTOR USING THE CHI-SQUARE TEST
Harvey Dillon Director, National Acoustic Laboratories Parliamentary Breakfast Hearing Awareness Week, 2012 Hearing aids – how much do they really help?
ECE 8443 – Pattern Recognition ECE 8423 – Adaptive Signal Processing Objectives: Introduction SNR Gain Patterns Beam Steering Shading Resources: Wiki:
Dept. of Electrical Engineering, KU Leuven, Belgium
Compressed Sensing Based UWB System Peng Zhang Wireless Networking System Lab WiNSys.
Digital Audio Signal Processing Lecture-4: Noise Reduction Marc Moonen/Alexander Bertrand Dept. E.E./ESAT-STADIUS, KU Leuven
Physical and perceptual evaluation of the Interaural Wiener Filter algorithm Simon Doclo 1, Thomas J. Klasen 1, Tim van den Bogaert 2, Marc Moonen 1,
Eigenstructure Methods for Noise Covariance Estimation Olawoye Oyeyele AICIP Group Presentation April 29th, 2003.
Nico De Clercq Pieter Gijsenbergh Noise reduction in hearing aids: Generalised Sidelobe Canceller.
Heart Sound Background Noise Removal Haim Appleboim Biomedical Seminar February 2007.
Acoustic impulse response measurement using speech and music signals John Usher Barcelona Media – Innovation Centre | Av. Diagonal, 177, planta 9,
Blind speech dereverberation using multiple microphones Inseon JANG, Seungjin CHOI Intelligent Multimedia Lab Department of Computer Science and Engineering,
Multiuser Detection (MUD) Combined with array signal processing in current wireless communication environments Wed. 박사 3학기 구 정 회.
Applied Psychoacoustics Lecture: Binaural Hearing Jonas Braasch Jens Blauert.
Image Restoration using Iterative Wiener Filter --- ECE533 Project Report Jing Liu, Yan Wu.
Digital Audio Signal Processing Lecture-2: Microphone Array Processing - Fixed Beamforming - Marc Moonen Dept. E.E./ESAT-STADIUS, KU Leuven
Nico De Clercq Pieter Gijsenbergh.  Problem  Solutions  Single-channel approach  Multichannel approach  Our assignment Overview.
Noise reduction and binaural cue preservation of multi- microphone algorithms Simon Doclo, Tim van den Bogaert, Marc Moonen, Jan Wouters Dept. of Electrical.
Performance analysis of channel estimation and adaptive equalization in slow fading channel Chen Zhifeng Electrical and Computer Engineering University.
Design of a robust multi- microphone noise reduction algorithm for hearing instruments Simon Doclo 1, Ann Spriet 1,2, Marc Moonen 1, Jan Wouters 2 1 Dept.
Authors: Sriram Ganapathy, Samuel Thomas, and Hynek Hermansky Temporal envelope compensation for robust phoneme recognition using modulation spectrum.
‘Missing Data’ speech recognition in reverberant conditions using binaural interaction Sue Harding, Jon Barker and Guy J. Brown Speech and Hearing Research.
L INKWITZ L AB S e n s i b l e R e p r o d u c t i o n & R e c o r d i n g o f A u d i t o r y S c e n e s Hearing Spatial Detail in Stereo Recordings.
Figures for Chapter 14 Binaural and bilateral issues Dillon (2001) Hearing Aids.
Signal Processing Algorithms for Wireless Acoustic Sensor Networks Alexander Bertrand Electrical Engineering Department (ESAT) Katholieke Universiteit.
Independent Component Analysis Algorithm for Adaptive Noise Cancelling 적응 잡음 제거를 위한 독립 성분 분석 알고리즘 Hyung-Min Park, Sang-Hoon Oh, and Soo-Young Lee Brain.
TIME-SHIFTED PRINCIPAL COMPONENT ANALYSIS BASED CUE EXTRACTION FOR STEREO AUDIO SIGNALS Jianjun HE, Ee-Leng Tan, Woon-Seng Gan Digital Signal Processing.
Hearing Research Center
1 Blind Channel Identification and Equalization in Dense Wireless Sensor Networks with Distributed Transmissions Xiaohua (Edward) Li Department of Electrical.
Study of Broadband Postbeamformer Interference Canceler Antenna Array Processor using Orthogonal Interference Beamformer Lal C. Godara and Presila Israt.
P. N. Kulkarni, P. C. Pandey, and D. S. Jangamashetti / DSP 2009, Santorini, 5-7 July DSP 2009 (Santorini, Greece. 5-7 July 2009), Session: S4P,
Simon Doclo1, Ann Spriet1,2, Marc Moonen1, Jan Wouters2
Laboratory for Experimental ORL K.U.Leuven, Belgium Dept. of Electrotechn. Eng. ESAT/SISTA K.U.Leuven, Belgium Combining noise reduction and binaural cue.
The Restricted Matched Filter for Distributed Detection Charles Sestok and Alan Oppenheim MIT DARPA SensIT PI Meeting Jan. 16, 2002.
Design and low-cost implementation of a robust multichannel noise reduction scheme for cochlear implants Simon Doclo 1, Ann Spriet 1,2, Jean-Baptiste Maj.
Digital Audio Signal Processing Lecture-3 Noise Reduction
Abstract: In many scenarios, wireless presents a tempting "last-mile" alternative to a wired connection for the delivery of internet service. However,
Evaluation of a Binaural FMV Beamforming Algorithm in Noise Jeffery B. Larsen, Charissa R. Lansing, Robert C. Bilger, Bruce Wheeler, Sandeep Phatak, Nandini.
Project-Final Presentation Blind Dereverberation Algorithm for Speech Signals Based on Multi-channel Linear Prediction Supervisor: Alexander Bertrand Authors:
Spatial Covariance Models For Under- Determined Reverberant Audio Source Separation N. Duong, E. Vincent and R. Gribonval METISS project team, IRISA/INRIA,
Motorola presents in collaboration with CNEL Introduction  Motivation: The limitation of traditional narrowband transmission channel  Advantage: Phone.
UNIT-IV. Introduction Speech signal is generated from a system. Generation is via excitation of system. Speech travels through various media. Nature of.
Compensating cocktail party noise with binaural spatial segregation on a novel device targeting partial hearing loss Luca Giuliani 1, Sara Sansalone 2,
Speech Enhancement Summer 2009
Auditory Localization in Rooms: Acoustic Analysis and Behavior
Precedence-based speech segregation in a virtual auditory environment
Liverpool Keele Contribution.
Fondazione Istituto Italiano di Tecnologia, Genoa, Italy
Antenna selection and RF processing for MIMO systems
Historic Document Image De-Noising using Principal Component Analysis (PCA) and Local Pixel Grouping (LPG) Han-Yang Tang1, Azah Kamilah Muda1, Yun-Huoy.
Analysis of Adaptive Array Algorithm Performance for Satellite Interference Cancellation in Radio Astronomy Lisha Li, Brian D. Jeffs, Andrew Poulsen, and.
Hearing Spatial Detail
Chen Zhifeng Electrical and Computer Engineering University of Florida
Chenhui Zheng/Communication Laboratory
Presentation transcript:

Reduced-bandwidth and distributed MWF-based noise reduction algorithms Simon Doclo, Tim Van den Bogaert, Jan Wouters, Marc Moonen Dept. of Electrical Engineering (ESAT-SCD), KU Leuven, Belgium Laboratory for Exp. ORL, KU Leuven, Belgium WASPAA-2007, Oct

2 Outline Hearing aids: bilateral vs. binaural processing Binaural multi-channel Wiener filter: transmit all microphone signals  large bandwidth of wireless link Reduce bandwidth: transmit only one contralateral signal osignal-independent: contralateral microphone, fixed beamformer osignal-dependent: MWF on contralateral microphones oiterative distributed MWF procedure: – rank-1 speech correlation matrix  converges to B-MWF solution ! – can still be used in practice when assumption is not satisfied Performance comparison: oSNR improvement (+ spatial directivity pattern) odB-MWF performance approaches quite well binaural MWF performance for all conditions

3 Many hearing impaired are fitted with hearing aid at both ears: oSignal processing to reduce background noise and improve speech intelligibility oSignal processing to preserve directional hearing (ILD/ITD cues) oMultiple microphone available: spectral + spatial processing IPD/ITD ILD Hearing aids: bilateral vs. binaural  Bilateral/binaural  Binaural MWF  Bandwidth reduction  Experimental results  Conclusions

4 Hearing aids: bilateral vs. binaural Bilateral system Independent left/right processing: binaural cues for localisation are distorted Binaural system - Larger SNR improvement (more microphones) - Preservation of binaural cues possible Need for binaural link  Bilateral/binaural  Binaural MWF  Bandwidth reduction  Experimental results  Conclusions

5 Hearing aids: bilateral vs. binaural Binaural multi-microphone noise reduction techniques: oFixed beamforming – Low complexity, but limited performance oAdaptive beamforming – Mostly based on GSC structure + e.g. passing low-pass portion unaltered to preserve ITD cues oComputational auditory scene analysis – Computation of (real-valued) binaural mask based on binaural and temporal/spectral cues oMulti-channel Wiener filtering – MMSE-based estimate of speech component in both hearing aids – Extensions for preserving binaural cues of speech and noise components [Desloge 1997, Merks 1997, Lotter 2006] [Welker 1997, Nishimura 2002, Lockwood 2004] [Kollmeier 1993, Wittkop 2003, Hamacher 2002, Haykin 2004] [Doclo, Klasen, Van den Bogaert, Wouters, Moonen ]  Bilateral/binaural  Binaural MWF  Bandwidth reduction  Experimental results  Conclusions

6 Configuration and notation M microphones on each hearing aid: Y 0, Y 1 Speech and noise components: Single speech source: (acoustic transfer functions) Collaboration: 2N signals transmitted between hearing aids  Bilateral/binaural  Binaural MWF  Bandwidth reduction  Experimental results  Conclusions

7 Binaural MWF (B-MWF) SDW-MWF using all 2M microphones from both hearing aids: oAll microphone signals are transmitted: oMMSE estimate of speech component in (front) microphone of left and right hearing aid + trade-off (  ) noise reductionspeech distortion speech component in front microphone Binaural MWF cost function: Estimated during speech-and-noise and noise-only periods: VAD  Bilateral/binaural  Binaural MWF  Bandwidth reduction  Experimental results  Conclusions

8 Binaural MWF (B-MWF) Optimal filters (general case): Optimal filters (single speech source): o is complex conjugate of speech ITF oOptimal filters at left and right hearing aid are parallel  Bilateral/binaural  Binaural MWF  Bandwidth reduction  Experimental results  Conclusions

9 To limit power/bandwidth requirements, transmit N=1 signal from contralateral hearing aid oB-MWF can still be obtained, namely if F 01 is parallel to and F 10 is parallel to  infeasible at first sight since full correlation matrices can not be computed ! Reduced-bandwidth algorithms  Bilateral/binaural  Binaural MWF  Bandwidth reduction -fixed beamformer -contralateral MWF -distributed scheme  Experimental results  Conclusions

10 Fixed beamformer Filters F 01 and F 10, which can be viewed as monaural beamformers, are signal-independent MWF-front: front contralateral microphone signals MWF-superd: monaural superdirective beamformer limited performance  Bilateral/binaural  Binaural MWF  Bandwidth reduction -fixed beamformer -contralateral MWF -distributed scheme  Experimental results  Conclusions

11 Contralateral MWF Transmitted signals = output of monaural MWF, estimating the contralateral speech component only using the contralateral microphone signals oSignal-dependent (better performance than signal-independent) oIncreased computational complexity (two MWF solutions for each hearing aid) In general suboptimal solution: oOptimal solution is obtained in case of single speech source and when noise components between left and right hearing aid are uncorrelated (unrealistic)  Bilateral/binaural  Binaural MWF  Bandwidth reduction -fixed beamformer -contralateral MWF -distributed scheme  Experimental results  Conclusions

12 Distributed MWF (dB-MWF) Iterative procedure: oIn each iteration F 10 is equal to W 00 from previous iteration, and F 01 is equal to W 11 from previous iteration  Bilateral/binaural  Binaural MWF  Bandwidth reduction -fixed beamformer -contralateral MWF -distributed scheme  Experimental results  Conclusions

13 Distributed MWF (dB-MWF)  Bilateral/binaural  Binaural MWF  Bandwidth reduction -fixed beamformer -contralateral MWF -distributed scheme  Experimental results  Conclusions

14 Distributed MWF (dB-MWF) Single speech source: convergence to B-MWF solution (!) oMWF cost function decreases in each step of iteration oConvergence to B-MWF solution, since it minimises J(W) AND satisfies with General case where R x is not a rank-1 matrix: oMWF cost function does not necessarily decrease in each iteration ousually no convergence to optimal B-MWF solution oAlthough, dB-MWF procedure can be used in practice and approaches B-MWF performance  Bilateral/binaural  Binaural MWF  Bandwidth reduction -fixed beamformer -contralateral MWF -distributed scheme  Experimental results  Conclusions

15 Experimental results Setup: oBinaural system with 2 omni microphones on each hearing aid, mounted on CORTEX MK2 artifical head in reverberant room  Bilateral/binaural  Binaural MWF  Bandwidth reduction  Experimental results -SNR improvement -directivity pattern  Conclusions oHRTFs: T 60  500 ms (and T 60  140 ms), f s = 20.48kHz oConfigurations: – speech source at 0  and several noise configurations (single, two and four noise sources) – speech source at 90  and noise source at 180  ospeech material = HINT, noise material = Auditec babble noise oInput SNR defined on LF microphone = 0dB (broadband) oIntelligibility-weighted SNR improvement between output signal and front microphone (L+R) MWF processing: oFrequency-domain batch procedure oL = 128,  =5 oPerfect VAD, odB-MWF procedure: K=10,

16 SNR improvement (500 ms - left HA) Original signal

17 B-MWF: oIn general largest SNR improvement of all algorithms oUp to 4 dB better than MWF-front (3 vs. 4 microphones) MWF-superd: oPerformance between MWF-front and B-MWF, but in general worse than (signal-dependent) MWF-contra and dB-MWF orelatively better performance when (signal-independent) directivity pattern of superdirective beamformer approaches optimal (signal- dependent) directivity pattern of B-MWF, e.g.  v =300  (left HA) MWF-contra: oPerformance between MWF-front and B-MWF dB-MWF: oBest performance of all reduced-bandwidth algorithms oSubstantial performance benefit compared to MWF-contra, especially for multiple noise sources oPerformance of dB-MWF approaches quite well performance of B-MWF, even though speech correlation matrices are not rank-1 due to FFT overlap and estimation errors, i.e. Experimental results  Bilateral/binaural  Binaural MWF  Bandwidth reduction  Experimental results -SNR improvement -directivity pattern  Conclusions

18 Experimental results Directivity pattern: oFullband spatial directivity pattern of F 01, i.e. the pattern generated using the right microphone signals and transmitted to the left hearing aid oConfiguration  v =[-120  120  ], T 60 = 140 ms oB-MWF: null steered towards direction of noise sources  optimally signal with high SNR should be transmitted oMWF-front, MWF-superd: directivity pattern not similar to B-MWF directivity pattern  low SNR improvement oMWF-contra: directivity pattern similar to B-MWF directivity pattern  high SNR improvement odB-MWF: best performance since directivity pattern closely matches B-MWF directivity pattern Using these spatial directivity patterns, it is possible to explain the performance of the different algorithms for different noise configurations to some extent  Bilateral/binaural  Binaural MWF  Bandwidth reduction  Experimental results -SNR improvement -directivity pattern  Conclusions

19 Contralateral directivity patterns (140 ms) B-MWFMWF-frontMWF-superd MWF-contradB-MWF  v =[-120  120  ]

20 Conclusions Binaural MWF: large bandwidth/power requirement Reduced-bandwidth algorithms: oMWF-front, MWF-superd: signal-independent oMWF-contra: monaural MWF using contralateral microphones – Signal-dependent, but suboptimal odB-MWF: iterative procedure – Converges to B-MWF solution for rank-1 speech correlation matrix – Also useful in practice when this assumption is not satisfied Experimental results: odB-MWF > MWF-contra > MWF-superd > MWF-front – Signal-dependent better than signal-independent – 2 or 3 iterations sufficient for dB-MWF procedure – dB-MWF performance approaches quite well B-MWF performance Extension: distributed processing in acoustic sensor networks  Bilateral/binaural  Binaural MWF  Bandwidth reduction  Experimental results  Conclusions