Effective speaker adaptations for speaker verification

Sungjoo Ahn, Sunmee Kang, Hanseok Ko

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

5 Citations (Scopus)

Abstract

This paper concerns effective speaker adaptation methods to solve the over-training problem in speaker verification, which frequently occurs when modeling a speaker with sparse training data. While various speaker adaptations have already been applied to speech recognition, these methods have not yet been formally considered in speaker verification. This paper proposes speaker adaptation methods using a combination of MAP and MLLR adaptations, which are successfully used in speech recognition, and applies to speaker verification. Our aim is to remedy the small training data problem by investigating effective speaker adaptations for speaker modeling. Experimental results show that the speaker verification system using a weighted MAP and MLLR adaptation outperforms that of the conventional speaker models without adaptation by a factor of up to 5 times. From these results, we show that the speaker adaptation method achieves significantly better performance even when only small training data is available for speaker verification.

Original languageEnglish
Title of host publicationICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
PublisherIEEE
Pages1081-1084
Number of pages4
Volume2
Publication statusPublished - 2000
Event2000 IEEE Interntional Conference on Acoustics, Speech, and Signal Processing - Istanbul, Turkey
Duration: 2000 Jun 52000 Jun 9

Other

Other2000 IEEE Interntional Conference on Acoustics, Speech, and Signal Processing
CityIstanbul, Turkey
Period00/6/500/6/9

Fingerprint

education
Speech recognition
speech recognition

ASJC Scopus subject areas

  • Signal Processing
  • Electrical and Electronic Engineering
  • Acoustics and Ultrasonics

Cite this

Ahn, S., Kang, S., & Ko, H. (2000). Effective speaker adaptations for speaker verification. In ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings (Vol. 2, pp. 1081-1084). IEEE.

Effective speaker adaptations for speaker verification. / Ahn, Sungjoo; Kang, Sunmee; Ko, Hanseok.

ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. Vol. 2 IEEE, 2000. p. 1081-1084.

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

Ahn, S, Kang, S & Ko, H 2000, Effective speaker adaptations for speaker verification. in ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. vol. 2, IEEE, pp. 1081-1084, 2000 IEEE Interntional Conference on Acoustics, Speech, and Signal Processing, Istanbul, Turkey, 00/6/5.
Ahn S, Kang S, Ko H. Effective speaker adaptations for speaker verification. In ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. Vol. 2. IEEE. 2000. p. 1081-1084
Ahn, Sungjoo ; Kang, Sunmee ; Ko, Hanseok. / Effective speaker adaptations for speaker verification. ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings. Vol. 2 IEEE, 2000. pp. 1081-1084
@inproceedings{4914c7e2d0d741ae85f5b709602d14e4,
title = "Effective speaker adaptations for speaker verification",
abstract = "This paper concerns effective speaker adaptation methods to solve the over-training problem in speaker verification, which frequently occurs when modeling a speaker with sparse training data. While various speaker adaptations have already been applied to speech recognition, these methods have not yet been formally considered in speaker verification. This paper proposes speaker adaptation methods using a combination of MAP and MLLR adaptations, which are successfully used in speech recognition, and applies to speaker verification. Our aim is to remedy the small training data problem by investigating effective speaker adaptations for speaker modeling. Experimental results show that the speaker verification system using a weighted MAP and MLLR adaptation outperforms that of the conventional speaker models without adaptation by a factor of up to 5 times. From these results, we show that the speaker adaptation method achieves significantly better performance even when only small training data is available for speaker verification.",
author = "Sungjoo Ahn and Sunmee Kang and Hanseok Ko",
year = "2000",
language = "English",
volume = "2",
pages = "1081--1084",
booktitle = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",
publisher = "IEEE",

}

TY - GEN

T1 - Effective speaker adaptations for speaker verification

AU - Ahn, Sungjoo

AU - Kang, Sunmee

AU - Ko, Hanseok

PY - 2000

Y1 - 2000

N2 - This paper concerns effective speaker adaptation methods to solve the over-training problem in speaker verification, which frequently occurs when modeling a speaker with sparse training data. While various speaker adaptations have already been applied to speech recognition, these methods have not yet been formally considered in speaker verification. This paper proposes speaker adaptation methods using a combination of MAP and MLLR adaptations, which are successfully used in speech recognition, and applies to speaker verification. Our aim is to remedy the small training data problem by investigating effective speaker adaptations for speaker modeling. Experimental results show that the speaker verification system using a weighted MAP and MLLR adaptation outperforms that of the conventional speaker models without adaptation by a factor of up to 5 times. From these results, we show that the speaker adaptation method achieves significantly better performance even when only small training data is available for speaker verification.

AB - This paper concerns effective speaker adaptation methods to solve the over-training problem in speaker verification, which frequently occurs when modeling a speaker with sparse training data. While various speaker adaptations have already been applied to speech recognition, these methods have not yet been formally considered in speaker verification. This paper proposes speaker adaptation methods using a combination of MAP and MLLR adaptations, which are successfully used in speech recognition, and applies to speaker verification. Our aim is to remedy the small training data problem by investigating effective speaker adaptations for speaker modeling. Experimental results show that the speaker verification system using a weighted MAP and MLLR adaptation outperforms that of the conventional speaker models without adaptation by a factor of up to 5 times. From these results, we show that the speaker adaptation method achieves significantly better performance even when only small training data is available for speaker verification.

UR - http://www.scopus.com/inward/record.url?scp=0033676899&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0033676899&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:0033676899

VL - 2

SP - 1081

EP - 1084

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

PB - IEEE

ER -