Complex extension of infinite sparse factor analysis for blind speech separation

Kohei Nagira, Toru Takahashi, Tetsuya Ogata, Hiroshi G. Okuno

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

We present a method of blind source separation (BSS) for speech signals using a complex extension of infinite sparse factor analysis (ISFA) in the frequency domain. Our method is robust against delayed signals that usually occur in real environments, such as reflections, short-time reverberations, and time lags of signals arriving at microphones. ISFA is a conventional non-parametric Bayesian method of BSS, which has only been applied to time domain signals because it can only deal with real signals. Our method uses complex normal distributions to estimate source signals and mixing matrix. Experimental results indicate that our method outperforms the conventional ISFA in the average signal-to-distortion ratio (SDR).

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages388-396
Number of pages9
Volume7191 LNCS
DOIs
Publication statusPublished - 2012
Externally publishedYes
Event10th International Conference on Latent Variable Analysis and Signal Separation, LVA/ICA 2012 - Tel Aviv
Duration: 2012 Mar 122012 Mar 15

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7191 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other10th International Conference on Latent Variable Analysis and Signal Separation, LVA/ICA 2012
CityTel Aviv
Period12/3/1212/3/15

Fingerprint

Factor analysis
Factor Analysis
Blind source separation
Blind Source Separation
Reverberation
Normal distribution
Microphones
Time Lag
Speech Signal
Nonparametric Methods
Bayesian Methods
Frequency Domain
Gaussian distribution
Speech
Time Domain
Experimental Results
Estimate

Keywords

  • Blind source separation
  • Infinite sparse factor analysis
  • Non-parametric Bayes

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Nagira, K., Takahashi, T., Ogata, T., & Okuno, H. G. (2012). Complex extension of infinite sparse factor analysis for blind speech separation. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7191 LNCS, pp. 388-396). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7191 LNCS). https://doi.org/10.1007/978-3-642-28551-6_48

Complex extension of infinite sparse factor analysis for blind speech separation. / Nagira, Kohei; Takahashi, Toru; Ogata, Tetsuya; Okuno, Hiroshi G.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 7191 LNCS 2012. p. 388-396 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7191 LNCS).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Nagira, K, Takahashi, T, Ogata, T & Okuno, HG 2012, Complex extension of infinite sparse factor analysis for blind speech separation. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 7191 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 7191 LNCS, pp. 388-396, 10th International Conference on Latent Variable Analysis and Signal Separation, LVA/ICA 2012, Tel Aviv, 12/3/12. https://doi.org/10.1007/978-3-642-28551-6_48
Nagira K, Takahashi T, Ogata T, Okuno HG. Complex extension of infinite sparse factor analysis for blind speech separation. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 7191 LNCS. 2012. p. 388-396. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-28551-6_48
Nagira, Kohei ; Takahashi, Toru ; Ogata, Tetsuya ; Okuno, Hiroshi G. / Complex extension of infinite sparse factor analysis for blind speech separation. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 7191 LNCS 2012. pp. 388-396 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{64f46d9fb4474744bbb573c9ff83a596,
title = "Complex extension of infinite sparse factor analysis for blind speech separation",
abstract = "We present a method of blind source separation (BSS) for speech signals using a complex extension of infinite sparse factor analysis (ISFA) in the frequency domain. Our method is robust against delayed signals that usually occur in real environments, such as reflections, short-time reverberations, and time lags of signals arriving at microphones. ISFA is a conventional non-parametric Bayesian method of BSS, which has only been applied to time domain signals because it can only deal with real signals. Our method uses complex normal distributions to estimate source signals and mixing matrix. Experimental results indicate that our method outperforms the conventional ISFA in the average signal-to-distortion ratio (SDR).",
keywords = "Blind source separation, Infinite sparse factor analysis, Non-parametric Bayes",
author = "Kohei Nagira and Toru Takahashi and Tetsuya Ogata and Okuno, {Hiroshi G.}",
year = "2012",
doi = "10.1007/978-3-642-28551-6_48",
language = "English",
isbn = "9783642285509",
volume = "7191 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "388--396",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Complex extension of infinite sparse factor analysis for blind speech separation

AU - Nagira, Kohei

AU - Takahashi, Toru

AU - Ogata, Tetsuya

AU - Okuno, Hiroshi G.

PY - 2012

Y1 - 2012

N2 - We present a method of blind source separation (BSS) for speech signals using a complex extension of infinite sparse factor analysis (ISFA) in the frequency domain. Our method is robust against delayed signals that usually occur in real environments, such as reflections, short-time reverberations, and time lags of signals arriving at microphones. ISFA is a conventional non-parametric Bayesian method of BSS, which has only been applied to time domain signals because it can only deal with real signals. Our method uses complex normal distributions to estimate source signals and mixing matrix. Experimental results indicate that our method outperforms the conventional ISFA in the average signal-to-distortion ratio (SDR).

AB - We present a method of blind source separation (BSS) for speech signals using a complex extension of infinite sparse factor analysis (ISFA) in the frequency domain. Our method is robust against delayed signals that usually occur in real environments, such as reflections, short-time reverberations, and time lags of signals arriving at microphones. ISFA is a conventional non-parametric Bayesian method of BSS, which has only been applied to time domain signals because it can only deal with real signals. Our method uses complex normal distributions to estimate source signals and mixing matrix. Experimental results indicate that our method outperforms the conventional ISFA in the average signal-to-distortion ratio (SDR).

KW - Blind source separation

KW - Infinite sparse factor analysis

KW - Non-parametric Bayes

UR - http://www.scopus.com/inward/record.url?scp=84863178508&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84863178508&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-28551-6_48

DO - 10.1007/978-3-642-28551-6_48

M3 - Conference contribution

AN - SCOPUS:84863178508

SN - 9783642285509

VL - 7191 LNCS

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

SP - 388

EP - 396

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

ER -