REGRESSION CLASS SELECTION AND SPEAKER ADAPTATION WITH MLLR IN MANDARIN CONTINUOUS SPEECH RECOGNITION

Chengrong Li; Jingdong Chen; Bo Xu

REGRESSION CLASS SELECTION AND SPEAKER ADAPTATION WITH MLLR IN MANDARIN CONTINUOUS SPEECH RECOGNITION

Chengrong Li, Jingdong Chen, Bo Xu

CAS - Institute of Automation

Research output: Contribution to conference › Paper › peer-review

2 Scopus citations

Abstract

Currently, CDHMM based continuous speech recognition has been widely extended to speaker-independent (SI) system. However, the performance of the SI system is highly dependent on the speakers, especially for Mandarin speech with accent, speaker adaptation becomes crucial important for real application. In this paper, MLLR approach is studied for speaker adaptation in mandarin continuous speech recognition and three approaches for defining regression classes are investigated: the first is based on Chinese phonetic classification, the second is based on statistical information of mixture distribution parameters and the third is based on state duration using segmental information. Other experiments like the effect of adaptation data and mixtures are presented also in the paper. The new variance-based regression class selecting scheme is proposed and has been proved to be effective.

Original language	English
Pages	2503-2506
Number of pages	4
State	Published - 1999
Externally published	Yes
Event	6th European Conference on Speech Communication and Technology, EUROSPEECH 1999 - Budapest, Hungary Duration: 5 Sep 1999 → 9 Sep 1999

Conference

Conference	6th European Conference on Speech Communication and Technology, EUROSPEECH 1999
Country/Territory	Hungary
City	Budapest
Period	5/09/99 → 9/09/99

Keywords

Mandarin speech recognition
MLLR
regression class
speaker adaptation

Cite this

@conference{c8ce529ba7e14a818c3c460901d35f1a,

title = "REGRESSION CLASS SELECTION AND SPEAKER ADAPTATION WITH MLLR IN MANDARIN CONTINUOUS SPEECH RECOGNITION",

abstract = "Currently, CDHMM based continuous speech recognition has been widely extended to speaker-independent (SI) system. However, the performance of the SI system is highly dependent on the speakers, especially for Mandarin speech with accent, speaker adaptation becomes crucial important for real application. In this paper, MLLR approach is studied for speaker adaptation in mandarin continuous speech recognition and three approaches for defining regression classes are investigated: the first is based on Chinese phonetic classification, the second is based on statistical information of mixture distribution parameters and the third is based on state duration using segmental information. Other experiments like the effect of adaptation data and mixtures are presented also in the paper. The new variance-based regression class selecting scheme is proposed and has been proved to be effective.",

keywords = "Mandarin speech recognition, MLLR, regression class, speaker adaptation",

author = "Chengrong Li and Jingdong Chen and Bo Xu",

note = "Publisher Copyright: {\textcopyright} 1999 6th European Conference on Speech Communication and Technology, EUROSPEECH 1999. All rights reserved.; 6th European Conference on Speech Communication and Technology, EUROSPEECH 1999 ; Conference date: 05-09-1999 Through 09-09-1999",

year = "1999",

language = "英语",

pages = "2503--2506",

}

TY - CONF

T1 - REGRESSION CLASS SELECTION AND SPEAKER ADAPTATION WITH MLLR IN MANDARIN CONTINUOUS SPEECH RECOGNITION

AU - Li, Chengrong

AU - Chen, Jingdong

AU - Xu, Bo

PY - 1999

Y1 - 1999

N2 - Currently, CDHMM based continuous speech recognition has been widely extended to speaker-independent (SI) system. However, the performance of the SI system is highly dependent on the speakers, especially for Mandarin speech with accent, speaker adaptation becomes crucial important for real application. In this paper, MLLR approach is studied for speaker adaptation in mandarin continuous speech recognition and three approaches for defining regression classes are investigated: the first is based on Chinese phonetic classification, the second is based on statistical information of mixture distribution parameters and the third is based on state duration using segmental information. Other experiments like the effect of adaptation data and mixtures are presented also in the paper. The new variance-based regression class selecting scheme is proposed and has been proved to be effective.

AB - Currently, CDHMM based continuous speech recognition has been widely extended to speaker-independent (SI) system. However, the performance of the SI system is highly dependent on the speakers, especially for Mandarin speech with accent, speaker adaptation becomes crucial important for real application. In this paper, MLLR approach is studied for speaker adaptation in mandarin continuous speech recognition and three approaches for defining regression classes are investigated: the first is based on Chinese phonetic classification, the second is based on statistical information of mixture distribution parameters and the third is based on state duration using segmental information. Other experiments like the effect of adaptation data and mixtures are presented also in the paper. The new variance-based regression class selecting scheme is proposed and has been proved to be effective.

KW - Mandarin speech recognition

KW - MLLR

KW - regression class

KW - speaker adaptation

UR - http://www.scopus.com/inward/record.url?scp=85009092922&partnerID=8YFLogxK

M3 - 论文

AN - SCOPUS:85009092922

SP - 2503

EP - 2506

T2 - 6th European Conference on Speech Communication and Technology, EUROSPEECH 1999

Y2 - 5 September 1999 through 9 September 1999

ER -

REGRESSION CLASS SELECTION AND SPEAKER ADAPTATION WITH MLLR IN MANDARIN CONTINUOUS SPEECH RECOGNITION

Abstract

Conference

Keywords

Other files and links

Fingerprint

Cite this