TY - JOUR
T1 - An approach to blind source separation based on temporal structure of speech signals
AU - Murata, Noboru
AU - Ikeda, Shiro
AU - Ziehe, Andreas
N1 - Funding Information:
We thank Shun-ichi Amari, Klaus-Robert Müller, Andrew Cichocki, and Colin Fyfe for very useful discussions on this work. We are grateful to Te-Won Lee for a set of data and comments on the manuscript. We also thank Kota Takahashi, Paris Smaragdis, and Scott C. Douglas for comments on our work and demonstrations which can be found in the web contents. We appreciate Daishi Harada for thorough proofreading our manuscript. Finally, we would like to thank the reviewers for their helpful comments. A.Z. was partly funded by DFG contracts JA 379/51 and 379/71.
PY - 2001
Y1 - 2001
N2 - In this paper, we introduce a new technique for blind source separation of speech signals. We focus on the temporal structure of the signals. The idea is to apply the decorrelation method proposed by Molgedey and Schuster in the time-frequency domain. Since we are applying separation algorithm on each frequency separately, we have to solve the amplitude and permutation ambiguity properly to reconstruct the separated signals. For solving the amplitude ambiguity, we use the matrix inversion and for the permutation ambiguity, we introduce a method based on the temporal structure of speech signals. We show some results of experiments with both artificially controlled data and speech data recorded in the real environment.
AB - In this paper, we introduce a new technique for blind source separation of speech signals. We focus on the temporal structure of the signals. The idea is to apply the decorrelation method proposed by Molgedey and Schuster in the time-frequency domain. Since we are applying separation algorithm on each frequency separately, we have to solve the amplitude and permutation ambiguity properly to reconstruct the separated signals. For solving the amplitude ambiguity, we use the matrix inversion and for the permutation ambiguity, we introduce a method based on the temporal structure of speech signals. We show some results of experiments with both artificially controlled data and speech data recorded in the real environment.
KW - Blind source separation
KW - Convolutive mixtures
KW - Time-frequency domain
UR - http://www.scopus.com/inward/record.url?scp=0035659640&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=0035659640&partnerID=8YFLogxK
U2 - 10.1016/S0925-2312(00)00345-3
DO - 10.1016/S0925-2312(00)00345-3
M3 - Article
AN - SCOPUS:0035659640
SN - 0925-2312
VL - 41
SP - 1
EP - 24
JO - Neurocomputing
JF - Neurocomputing
IS - 1-4
ER -