HMM approach to text-prompted speaker verification

Chi Wei Che; Qiguang Lin; Dong Suk Yuk

HMM approach to text-prompted speaker verification

Chi Wei Che, Qiguang Lin, Dong Suk Yuk

Research output: Contribution to journal › Conference article › peer-review

Abstract

This paper presents a speaker recognition system based on Hidden Markov Modes (HMM). The system utilizes concatenated phoneme HMMs and works in a text-prompted mode. Each registered speaker has a separate set of HMMs which are trained using the Baum-Welch algorithm. The speaker recognition system has been evaluated with the YOHO voice verification corpus in terms of both speaker verification and closed-set speaker identification. It is shown that by using 10 seconds of testing speech, an error rate of 0.09% for male and 0.29% for female are obtained for speaker identification with a total population of 138 talkers. For speaker verification, under the 0% false rejection condition, the system achieves a false acceptance rate of 0.09% for male and 0% for female. This paper also studies effects of various factors (such as the mixture number and cohort selection) on the performance of speaker recognition.

Original language	English
Pages (from-to)	673-676
Number of pages	4
Journal	ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
Volume	2
Publication status	Published - 1996
Externally published	Yes
Event	Proceedings of the 1996 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP. Part 1 (of 6) - Atlanta, GA, USA Duration: 1996 May 7 → 1996 May 10

ASJC Scopus subject areas

Software
Signal Processing
Electrical and Electronic Engineering

Cite this

@article{1b73a447b7ea463c96137b1edd035ade,

title = "HMM approach to text-prompted speaker verification",

abstract = "This paper presents a speaker recognition system based on Hidden Markov Modes (HMM). The system utilizes concatenated phoneme HMMs and works in a text-prompted mode. Each registered speaker has a separate set of HMMs which are trained using the Baum-Welch algorithm. The speaker recognition system has been evaluated with the YOHO voice verification corpus in terms of both speaker verification and closed-set speaker identification. It is shown that by using 10 seconds of testing speech, an error rate of 0.09% for male and 0.29% for female are obtained for speaker identification with a total population of 138 talkers. For speaker verification, under the 0% false rejection condition, the system achieves a false acceptance rate of 0.09% for male and 0% for female. This paper also studies effects of various factors (such as the mixture number and cohort selection) on the performance of speaker recognition.",

author = "Che, {Chi Wei} and Qiguang Lin and Yuk, {Dong Suk}",

year = "1996",

language = "English",

volume = "2",

pages = "673--676",

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

issn = "0736-7791",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

note = "Proceedings of the 1996 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP. Part 1 (of 6) ; Conference date: 07-05-1996 Through 10-05-1996",

}

TY - JOUR

T1 - HMM approach to text-prompted speaker verification

AU - Che, Chi Wei

AU - Lin, Qiguang

AU - Yuk, Dong Suk

PY - 1996

Y1 - 1996

N2 - This paper presents a speaker recognition system based on Hidden Markov Modes (HMM). The system utilizes concatenated phoneme HMMs and works in a text-prompted mode. Each registered speaker has a separate set of HMMs which are trained using the Baum-Welch algorithm. The speaker recognition system has been evaluated with the YOHO voice verification corpus in terms of both speaker verification and closed-set speaker identification. It is shown that by using 10 seconds of testing speech, an error rate of 0.09% for male and 0.29% for female are obtained for speaker identification with a total population of 138 talkers. For speaker verification, under the 0% false rejection condition, the system achieves a false acceptance rate of 0.09% for male and 0% for female. This paper also studies effects of various factors (such as the mixture number and cohort selection) on the performance of speaker recognition.

AB - This paper presents a speaker recognition system based on Hidden Markov Modes (HMM). The system utilizes concatenated phoneme HMMs and works in a text-prompted mode. Each registered speaker has a separate set of HMMs which are trained using the Baum-Welch algorithm. The speaker recognition system has been evaluated with the YOHO voice verification corpus in terms of both speaker verification and closed-set speaker identification. It is shown that by using 10 seconds of testing speech, an error rate of 0.09% for male and 0.29% for female are obtained for speaker identification with a total population of 138 talkers. For speaker verification, under the 0% false rejection condition, the system achieves a false acceptance rate of 0.09% for male and 0% for female. This paper also studies effects of various factors (such as the mixture number and cohort selection) on the performance of speaker recognition.

UR - http://www.scopus.com/inward/record.url?scp=0029765817&partnerID=8YFLogxK

M3 - Conference article

AN - SCOPUS:0029765817

SN - 0736-7791

VL - 2

SP - 673

EP - 676

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

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

T2 - Proceedings of the 1996 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP. Part 1 (of 6)

Y2 - 7 May 1996 through 10 May 1996

ER -

HMM approach to text-prompted speaker verification

Abstract

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this