Incorporating MAP estimation and covariance transform for SVM based speaker recognition

Cheung Chi Leung; Donglai Zhu; Kong Aik Lee; Bin Ma; Haizhou Li

Incorporating MAP estimation and covariance transform for SVM based speaker recognition

Cheung Chi Leung, Donglai Zhu, Kong Aik Lee, Bin Ma, Haizhou Li

Research output: Chapter in book / Conference proceeding › Conference article published in proceeding or book › Academic research › peer-review

Abstract

In this paper, we apply Constrained Maximum a Posteriori Linear Regression (CMAPLR) transformation on Universal Background Model (UBM) when characterizing each speaker with a supervector. We incorporate the covariance transformation parameters into the supervector in addition to the mean transformation parameters. Maximum Likelihood Linear Regression (MLLR) covariance transformation is adopted. The auxiliary function maximization involved in Maximum Likelihood (ML) and Maximum a Posteriori (MAP) estimation is also presented. Our experiment on the 2006 NIST Speaker Recognition Evaluation (SRE) corpus shows that the two proposed techniques provide substantial performance improvement.

Original language	English
Title of host publication	Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010
Publisher	International Speech Communication Association
Pages	2318-2321
Number of pages	4
Publication status	Published - Sept 2010
Externally published	Yes

Publication series

Name	Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010

Keywords

MAPLR
MLLR
Speaker adaptation
Speaker recognition

ASJC Scopus subject areas

Language and Linguistics
Speech and Hearing
Human-Computer Interaction
Signal Processing
Software
Modelling and Simulation

Cite this

Leung, C. C., Zhu, D., Lee, K. A., Ma, B., & Li, H. (2010). Incorporating MAP estimation and covariance transform for SVM based speaker recognition. In Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010 (pp. 2318-2321). (Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010). International Speech Communication Association.

Leung, Cheung Chi ; Zhu, Donglai ; Lee, Kong Aik et al. / Incorporating MAP estimation and covariance transform for SVM based speaker recognition. Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010. International Speech Communication Association, 2010. pp. 2318-2321 (Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010).

@inproceedings{73eda7977c334f67bdc9574c835bddcf,

title = "Incorporating MAP estimation and covariance transform for SVM based speaker recognition",

abstract = "In this paper, we apply Constrained Maximum a Posteriori Linear Regression (CMAPLR) transformation on Universal Background Model (UBM) when characterizing each speaker with a supervector. We incorporate the covariance transformation parameters into the supervector in addition to the mean transformation parameters. Maximum Likelihood Linear Regression (MLLR) covariance transformation is adopted. The auxiliary function maximization involved in Maximum Likelihood (ML) and Maximum a Posteriori (MAP) estimation is also presented. Our experiment on the 2006 NIST Speaker Recognition Evaluation (SRE) corpus shows that the two proposed techniques provide substantial performance improvement.",

keywords = "MAPLR, MLLR, Speaker adaptation, Speaker recognition",

author = "Leung, {Cheung Chi} and Donglai Zhu and Lee, {Kong Aik} and Bin Ma and Haizhou Li",

year = "2010",

month = sep,

language = "English",

series = "Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010",

publisher = "International Speech Communication Association",

pages = "2318--2321",

booktitle = "Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010",

}

Leung, CC, Zhu, D, Lee, KA, Ma, B & Li, H 2010, Incorporating MAP estimation and covariance transform for SVM based speaker recognition. in Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010. Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010, International Speech Communication Association, pp. 2318-2321.

Incorporating MAP estimation and covariance transform for SVM based speaker recognition. / Leung, Cheung Chi; Zhu, Donglai; Lee, Kong Aik et al.
Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010. International Speech Communication Association, 2010. p. 2318-2321 (Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010).

Research output: Chapter in book / Conference proceeding › Conference article published in proceeding or book › Academic research › peer-review

TY - GEN

T1 - Incorporating MAP estimation and covariance transform for SVM based speaker recognition

AU - Leung, Cheung Chi

AU - Zhu, Donglai

AU - Lee, Kong Aik

AU - Ma, Bin

AU - Li, Haizhou

PY - 2010/9

Y1 - 2010/9

N2 - In this paper, we apply Constrained Maximum a Posteriori Linear Regression (CMAPLR) transformation on Universal Background Model (UBM) when characterizing each speaker with a supervector. We incorporate the covariance transformation parameters into the supervector in addition to the mean transformation parameters. Maximum Likelihood Linear Regression (MLLR) covariance transformation is adopted. The auxiliary function maximization involved in Maximum Likelihood (ML) and Maximum a Posteriori (MAP) estimation is also presented. Our experiment on the 2006 NIST Speaker Recognition Evaluation (SRE) corpus shows that the two proposed techniques provide substantial performance improvement.

AB - In this paper, we apply Constrained Maximum a Posteriori Linear Regression (CMAPLR) transformation on Universal Background Model (UBM) when characterizing each speaker with a supervector. We incorporate the covariance transformation parameters into the supervector in addition to the mean transformation parameters. Maximum Likelihood Linear Regression (MLLR) covariance transformation is adopted. The auxiliary function maximization involved in Maximum Likelihood (ML) and Maximum a Posteriori (MAP) estimation is also presented. Our experiment on the 2006 NIST Speaker Recognition Evaluation (SRE) corpus shows that the two proposed techniques provide substantial performance improvement.

KW - MAPLR

KW - MLLR

KW - Speaker adaptation

KW - Speaker recognition

UR - http://www.scopus.com/inward/record.url?scp=79959825356&partnerID=8YFLogxK

M3 - Conference article published in proceeding or book

AN - SCOPUS:79959825356

T3 - Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010

SP - 2318

EP - 2321

BT - Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010

PB - International Speech Communication Association

ER -

Leung CC, Zhu D, Lee KA, Ma B, Li H. Incorporating MAP estimation and covariance transform for SVM based speaker recognition. In Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010. International Speech Communication Association. 2010. p. 2318-2321. (Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010).

Incorporating MAP estimation and covariance transform for SVM based speaker recognition

Abstract

Publication series

Keywords

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this