Estimating the number of sources for frequency-domain blind source separation

Hiroshi Sawada*, Stefan Winter, Ryo Mukai, Shoko Araki, Shoji Makino

*この研究の対応する著者

研究成果: Chapter

9 被引用数 (Scopus)

抄録

Blind source separation (BSS) for convolutive mixtures can be performed efficiently in the frequency domain, where independent component analysis (ICA) is applied separately in each frequency bin. To solve the permutation problem of frequency-domain BSS robustly, information regarding the number of sources is very important. This paper presents a method for estimating the number of sources from convolutive mixtures of sources. The new method estimates the power of each source or noise component by using ICA and a scaling technique to distinguish sources and noises. Also, a reverberant component can be identified by calculating the correlation of component envelopes. Experimental results for up to three sources show that the proposed method worked well in a reverberant condition whose reverberation time was 200 ms.

本文言語English
ホスト出版物のタイトルLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
編集者Carlos G. Puntonet, Alberto Prieto
出版社Springer Verlag
ページ610-617
ページ数8
ISBN(電子版)3540230564, 9783540230564
DOI
出版ステータスPublished - 2004
外部発表はい

出版物シリーズ

名前Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
3195
ISSN(印刷版)0302-9743
ISSN(電子版)1611-3349

ASJC Scopus subject areas

  • 理論的コンピュータサイエンス
  • コンピュータサイエンス一般

フィンガープリント

「Estimating the number of sources for frequency-domain blind source separation」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。

引用スタイル