Blind Source Separation of Convolutive Mixtures of Speech in Frequency Domain

Shoji Makino, Hiroshi Sawada, Ryo Mukai, Shoko ArakiIEICE Trans. Fundam. Electron. Commun. Comput. Sci 2005 · OpenReview

Abstract

This paper overviews a total solution for frequency-domain blind source separation (BSS) of convolutive mixtures of audio signals, especially speech. Frequency-domain BSS performs independent component analysis (ICA) in each frequency bin, and this is more efficient than time-domain BSS. We describe a sophisticated total solution for frequency-domain BSS, including permutation, scaling, circularity, and complex activation function solutions. Experimental results of 2<cd0215f.gif>2, 3<cd0215f.gif>3, 4<cd0215f.gif>4, 6<cd0215f.gif>8, and 2<cd0215f.gif>2 (moving sources), (#sources<cd0215f.gif>#microphones) in a room are promising.
BibTeX
@article{makino2005blind,
  author = {Shoji Makino and Hiroshi Sawada and Ryo Mukai and Shoko Araki},
  title = {{Blind Source Separation of Convolutive Mixtures of Speech in Frequency Domain}},
  journal = {IEICE Trans. Fundam. Electron. Commun. Comput. Sci.},
  year = {2005},
  url = {https://openreview.net/forum?id=DENQUVV6pU},
}

Loading related content...

Figure