Underdetermined blind separation of convolutive mixtures of speech with directivity pattern based mask and ICA

Shoko Araki; Shoji Makino; Hiroshi Sawada; Ryo Mukai

doi:10.1007/978-3-540-30110-3_113

Underdetermined blind separation of convolutive mixtures of speech with directivity pattern based mask and ICA

Shoko Araki^*, Shoji Makino, Hiroshi Sawada, Ryo Mukai

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Chapter

16 Citations (Scopus)

Abstract

We propose a method for separating N speech signals with M sensors where N > M. Some existing methods employ binary masks to extract the signals, and therefore, the extracted signals contain loud musical noise. To overcome this problem, we propose using a directivity pattern based continuous mask, which masks N - M sources in the observations, and independent component analysis (ICA) to separate the remaining mixtures. We conducted experiments for N = 3 with M = 2 and N = 4 with M = 2, and obtained separated signals with little distortion.

Original language	English
Title of host publication	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Editors	Carlos G. Puntonet, Alberto Prieto
Publisher	Springer Verlag
Pages	898-905
Number of pages	8
ISBN (Electronic)	3540230564, 9783540230564
DOIs	https://doi.org/10.1007/978-3-540-30110-3_113
Publication status	Published - 2004
Externally published	Yes

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	3195
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

ASJC Scopus subject areas

Theoretical Computer Science
Computer Science(all)

Access to Document

10.1007/978-3-540-30110-3_113

Cite this

Araki, S., Makino, S., Sawada, H., & Mukai, R. (2004). Underdetermined blind separation of convolutive mixtures of speech with directivity pattern based mask and ICA. In C. G. Puntonet, & A. Prieto (Eds.), Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (pp. 898-905). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3195). Springer Verlag. https://doi.org/10.1007/978-3-540-30110-3_113

Underdetermined blind separation of convolutive mixtures of speech with directivity pattern based mask and ICA. / Araki, Shoko; Makino, Shoji; Sawada, Hiroshi et al.
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). ed. / Carlos G. Puntonet; Alberto Prieto. Springer Verlag, 2004. p. 898-905 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 3195).

Research output: Chapter in Book/Report/Conference proceeding › Chapter

Araki, S, Makino, S, Sawada, H & Mukai, R 2004, Underdetermined blind separation of convolutive mixtures of speech with directivity pattern based mask and ICA. in CG Puntonet & A Prieto (eds), Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 3195, Springer Verlag, pp. 898-905. https://doi.org/10.1007/978-3-540-30110-3_113

Araki S, Makino S, Sawada H, Mukai R. Underdetermined blind separation of convolutive mixtures of speech with directivity pattern based mask and ICA. In Puntonet CG, Prieto A, editors, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Springer Verlag. 2004. p. 898-905. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-540-30110-3_113

Araki, Shoko ; Makino, Shoji ; Sawada, Hiroshi et al. / Underdetermined blind separation of convolutive mixtures of speech with directivity pattern based mask and ICA. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). editor / Carlos G. Puntonet ; Alberto Prieto. Springer Verlag, 2004. pp. 898-905 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inbook{019ecb809ff04ec4ad48411c94778baa,

title = "Underdetermined blind separation of convolutive mixtures of speech with directivity pattern based mask and ICA",

abstract = "We propose a method for separating N speech signals with M sensors where N > M. Some existing methods employ binary masks to extract the signals, and therefore, the extracted signals contain loud musical noise. To overcome this problem, we propose using a directivity pattern based continuous mask, which masks N - M sources in the observations, and independent component analysis (ICA) to separate the remaining mixtures. We conducted experiments for N = 3 with M = 2 and N = 4 with M = 2, and obtained separated signals with little distortion.",

author = "Shoko Araki and Shoji Makino and Hiroshi Sawada and Ryo Mukai",

year = "2004",

doi = "10.1007/978-3-540-30110-3_113",

language = "English",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "898--905",

editor = "Puntonet, {Carlos G.} and Alberto Prieto",

booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - CHAP

T1 - Underdetermined blind separation of convolutive mixtures of speech with directivity pattern based mask and ICA

AU - Araki, Shoko

AU - Makino, Shoji

AU - Sawada, Hiroshi

AU - Mukai, Ryo

PY - 2004

Y1 - 2004

N2 - We propose a method for separating N speech signals with M sensors where N > M. Some existing methods employ binary masks to extract the signals, and therefore, the extracted signals contain loud musical noise. To overcome this problem, we propose using a directivity pattern based continuous mask, which masks N - M sources in the observations, and independent component analysis (ICA) to separate the remaining mixtures. We conducted experiments for N = 3 with M = 2 and N = 4 with M = 2, and obtained separated signals with little distortion.

AB - We propose a method for separating N speech signals with M sensors where N > M. Some existing methods employ binary masks to extract the signals, and therefore, the extracted signals contain loud musical noise. To overcome this problem, we propose using a directivity pattern based continuous mask, which masks N - M sources in the observations, and independent component analysis (ICA) to separate the remaining mixtures. We conducted experiments for N = 3 with M = 2 and N = 4 with M = 2, and obtained separated signals with little distortion.

UR - http://www.scopus.com/inward/record.url?scp=35048881485&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=35048881485&partnerID=8YFLogxK

U2 - 10.1007/978-3-540-30110-3_113

DO - 10.1007/978-3-540-30110-3_113

M3 - Chapter

AN - SCOPUS:35048881485

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 898

EP - 905

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

A2 - Puntonet, Carlos G.

A2 - Prieto, Alberto

PB - Springer Verlag

ER -

Underdetermined blind separation of convolutive mixtures of speech with directivity pattern based mask and ICA

Abstract

Publication series

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this