Fully Bayesian inference of multi-mixture Gaussian model and its evaluation using speaker clustering

Naohiro Tawara; Tetsuji Ogawa; Shinji Watanabe; Tetsunori Kobayashi

doi:10.1109/ICASSP.2012.6289105

Fully Bayesian inference of multi-mixture Gaussian model and its evaluation using speaker clustering

Naohiro Tawara^*, Tetsuji Ogawa, Shinji Watanabe, Tetsunori Kobayashi

^*Corresponding author for this work

School of Fundamental Science and Engineering

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

6 Citations (Scopus)

Abstract

This study aims to verify effective optimization methods for estimating parametric, fully Bayesian models in speech processing. For that purpose, we investigate the impact of the difference in optimization methods for the multi-scale Gaussian mixture model, which is suitable for speaker clustering, on the clustering accuracy. The Markov chain Monte Carlo (MCMC)-based method was compared with the variational Bayesian method in the speaker clustering experiment; with a small amount of data, the MCMC-based method was more effective; with large scale data (more than one million samples), the difference between these methods in terms of the clustering accuracy decreased and the MCMC-based method was computationally efficient.

Original language	English
Title of host publication	2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings
Pages	5253-5256
Number of pages	4
DOIs	https://doi.org/10.1109/ICASSP.2012.6289105
Publication status	Published - 2012
Event	2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Kyoto, Japan Duration: 2012 Mar 25 → 2012 Mar 30

Publication series

Name	ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
ISSN (Print)	1520-6149

Conference

Conference	2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012
Country/Territory	Japan
City	Kyoto
Period	12/3/25 → 12/3/30

Keywords

Gibbs sampling
Speaker clustering
multi-scale Gaussian mixture model
variational Bayesian method

ASJC Scopus subject areas

Software
Signal Processing
Electrical and Electronic Engineering

Access to Document

10.1109/ICASSP.2012.6289105

Cite this

Tawara, N., Ogawa, T., Watanabe, S., & Kobayashi, T. (2012). Fully Bayesian inference of multi-mixture Gaussian model and its evaluation using speaker clustering. In 2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings (pp. 5253-5256). Article 6289105 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). https://doi.org/10.1109/ICASSP.2012.6289105

Fully Bayesian inference of multi-mixture Gaussian model and its evaluation using speaker clustering. / Tawara, Naohiro; Ogawa, Tetsuji ; Watanabe, Shinji et al.
2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings. 2012. p. 5253-5256 6289105 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings).

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

Tawara, N, Ogawa, T , Watanabe, S & Kobayashi, T 2012, Fully Bayesian inference of multi-mixture Gaussian model and its evaluation using speaker clustering. in 2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings., 6289105, ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, pp. 5253-5256, 2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012, Kyoto, Japan, 12/3/25. https://doi.org/10.1109/ICASSP.2012.6289105

Tawara N, Ogawa T , Watanabe S , Kobayashi T. Fully Bayesian inference of multi-mixture Gaussian model and its evaluation using speaker clustering. In 2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings. 2012. p. 5253-5256. 6289105. (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). doi: 10.1109/ICASSP.2012.6289105

Tawara, Naohiro ; Ogawa, Tetsuji ; Watanabe, Shinji et al. / Fully Bayesian inference of multi-mixture Gaussian model and its evaluation using speaker clustering. 2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings. 2012. pp. 5253-5256 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings).

@inproceedings{4b0cdebe575d491aa99957eb361960b2,

title = "Fully Bayesian inference of multi-mixture Gaussian model and its evaluation using speaker clustering",

abstract = "This study aims to verify effective optimization methods for estimating parametric, fully Bayesian models in speech processing. For that purpose, we investigate the impact of the difference in optimization methods for the multi-scale Gaussian mixture model, which is suitable for speaker clustering, on the clustering accuracy. The Markov chain Monte Carlo (MCMC)-based method was compared with the variational Bayesian method in the speaker clustering experiment; with a small amount of data, the MCMC-based method was more effective; with large scale data (more than one million samples), the difference between these methods in terms of the clustering accuracy decreased and the MCMC-based method was computationally efficient.",

keywords = "Gibbs sampling, Speaker clustering, multi-scale Gaussian mixture model, variational Bayesian method",

author = "Naohiro Tawara and Tetsuji Ogawa and Shinji Watanabe and Tetsunori Kobayashi",

year = "2012",

doi = "10.1109/ICASSP.2012.6289105",

language = "English",

isbn = "9781467300469",

series = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",

pages = "5253--5256",

booktitle = "2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings",

note = "2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 ; Conference date: 25-03-2012 Through 30-03-2012",

}

TY - GEN

T1 - Fully Bayesian inference of multi-mixture Gaussian model and its evaluation using speaker clustering

AU - Tawara, Naohiro

AU - Ogawa, Tetsuji

AU - Watanabe, Shinji

AU - Kobayashi, Tetsunori

PY - 2012

Y1 - 2012

N2 - This study aims to verify effective optimization methods for estimating parametric, fully Bayesian models in speech processing. For that purpose, we investigate the impact of the difference in optimization methods for the multi-scale Gaussian mixture model, which is suitable for speaker clustering, on the clustering accuracy. The Markov chain Monte Carlo (MCMC)-based method was compared with the variational Bayesian method in the speaker clustering experiment; with a small amount of data, the MCMC-based method was more effective; with large scale data (more than one million samples), the difference between these methods in terms of the clustering accuracy decreased and the MCMC-based method was computationally efficient.

AB - This study aims to verify effective optimization methods for estimating parametric, fully Bayesian models in speech processing. For that purpose, we investigate the impact of the difference in optimization methods for the multi-scale Gaussian mixture model, which is suitable for speaker clustering, on the clustering accuracy. The Markov chain Monte Carlo (MCMC)-based method was compared with the variational Bayesian method in the speaker clustering experiment; with a small amount of data, the MCMC-based method was more effective; with large scale data (more than one million samples), the difference between these methods in terms of the clustering accuracy decreased and the MCMC-based method was computationally efficient.

KW - Gibbs sampling

KW - Speaker clustering

KW - multi-scale Gaussian mixture model

KW - variational Bayesian method

UR - http://www.scopus.com/inward/record.url?scp=84867626020&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84867626020&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2012.6289105

DO - 10.1109/ICASSP.2012.6289105

M3 - Conference contribution

AN - SCOPUS:84867626020

SN - 9781467300469

T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

SP - 5253

EP - 5256

BT - 2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012 - Proceedings

T2 - 2012 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2012

Y2 - 25 March 2012 through 30 March 2012

ER -

Fully Bayesian inference of multi-mixture Gaussian model and its evaluation using speaker clustering

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this