DocumentCode :
2704270
Title :
Subband-Based Blind Signal Processing for Source Separation in Convolutive Mixtures of Speech
Author :
Kokkinakis, Kostas ; Loizou, Philipos C.
Author_Institution :
Dept. of Electr. Eng., Texas Univ., Dallas, TX
Volume :
4
fYear :
2007
fDate :
15-20 April 2007
Abstract :
This paper describes a highly practical blind signal separation (BSS) scheme operating on subband domain data to blindly segregate convolutive mixtures of speech. The proposed method relies on spatiotemporal separation carried out in the time domain by using a multichannel blind deconvolution (MBD) algorithm that enforces separation by entropy maximization through the popular natural gradient algorithm (NGA). Numerical experiments with binaural impulse responses affirm the validity and illustrate the practical appeal of the presented technique even for difficult speech separation setups.
Keywords :
blind source separation; maximum entropy methods; speech processing; time-domain analysis; transient response; binaural impulse responses; blind signal separation; convolutive speech mixtures; entropy maximization; multichannel blind deconvolution; natural gradient algorithm; source separation; spatiotemporal separation; subband-based blind signal processing; Blind source separation; Deconvolution; Discrete Fourier transforms; Finite impulse response filter; MIMO; Sensor systems; Signal processing; Signal processing algorithms; Source separation; Speech processing; Subband filtering; blind source separation; convolutive speech mixtures; multichannel blind deconvolution;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Acoustics, Speech and Signal Processing, 2007. ICASSP 2007. IEEE International Conference on
Conference_Location :
Honolulu, HI
ISSN :
1520-6149
Print_ISBN :
1-4244-0727-3
Electronic_ISBN :
1520-6149
Type :
conf
DOI :
10.1109/ICASSP.2007.367220
Filename :
4218251
Link To Document :
بازگشت