Advances in mandarin broadcast speech recognition

Mei Yuh Hwang; Wen Wang; Xin Lei; Jing Zheng; Ozgur Cetin; Gang Peng

Advances in mandarin broadcast speech recognition

Mei Yuh Hwang, Wen Wang, Xin Lei, Jing Zheng, Ozgur Cetin, Gang Peng

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

6 Citations (Scopus)

Abstract

We describe our continuing efforts to improve the UW-SRI-ICSI Mandarin broadcast speech recognizer. This includes increasing acoustic and text training data, adding discriminative features, incorporating frame-level discriminative training criterion, multiple-pass acoustic model (AM) cross adaptation, language model (LM) genre adaptation and system combination. The net effect without LM adaptation was a 24%-64% relative reduction in character error rates (CERs) on a variety of test sets. In addition, LM adaptation gave us another 6% of relative CER reduction on broadcast conversations.

Original language	English
Title of host publication	International Speech Communication Association - 8th Annual Conference of the International Speech Communication Association, Interspeech 2007
Pages	2876-2879
Number of pages	4
Volume	4
Publication status	Published - 1 Dec 2007
Externally published	Yes
Event	8th Annual Conference of the International Speech Communication Association, Interspeech 2007 - Antwerp, Belgium Duration: 27 Aug 2007 → 31 Aug 2007

Conference

Conference	8th Annual Conference of the International Speech Communication Association, Interspeech 2007
Country/Territory	Belgium
City	Antwerp
Period	27/08/07 → 31/08/07

Keywords

Character error rates
Discriminative features
Discriminative training
LM adaptation
Mandarin

ASJC Scopus subject areas

Computer Science Applications
Software
Modelling and Simulation
Linguistics and Language
Communication

Cite this

@inproceedings{f5dd2fc988aa46ac8fcf5ec4a1b66897,

title = "Advances in mandarin broadcast speech recognition",

abstract = "We describe our continuing efforts to improve the UW-SRI-ICSI Mandarin broadcast speech recognizer. This includes increasing acoustic and text training data, adding discriminative features, incorporating frame-level discriminative training criterion, multiple-pass acoustic model (AM) cross adaptation, language model (LM) genre adaptation and system combination. The net effect without LM adaptation was a 24%-64% relative reduction in character error rates (CERs) on a variety of test sets. In addition, LM adaptation gave us another 6% of relative CER reduction on broadcast conversations.",

keywords = "Character error rates, Discriminative features, Discriminative training, LM adaptation, Mandarin",

author = "Hwang, {Mei Yuh} and Wen Wang and Xin Lei and Jing Zheng and Ozgur Cetin and Gang Peng",

year = "2007",

month = dec,

day = "1",

language = "English",

isbn = "9781605603162",

volume = "4",

pages = "2876--2879",

booktitle = "International Speech Communication Association - 8th Annual Conference of the International Speech Communication Association, Interspeech 2007",

note = "8th Annual Conference of the International Speech Communication Association, Interspeech 2007 ; Conference date: 27-08-2007 Through 31-08-2007",

}

Hwang, MY, Wang, W, Lei, X, Zheng, J, Cetin, O & Peng, G 2007, Advances in mandarin broadcast speech recognition. in International Speech Communication Association - 8th Annual Conference of the International Speech Communication Association, Interspeech 2007. vol. 4, pp. 2876-2879, 8th Annual Conference of the International Speech Communication Association, Interspeech 2007, Antwerp, Belgium, 27/08/07.

Advances in mandarin broadcast speech recognition. / Hwang, Mei Yuh; Wang, Wen; Lei, Xin et al.
International Speech Communication Association - 8th Annual Conference of the International Speech Communication Association, Interspeech 2007. Vol. 4 2007. p. 2876-2879.

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

TY - GEN

T1 - Advances in mandarin broadcast speech recognition

AU - Hwang, Mei Yuh

AU - Wang, Wen

AU - Lei, Xin

AU - Zheng, Jing

AU - Cetin, Ozgur

AU - Peng, Gang

PY - 2007/12/1

Y1 - 2007/12/1

N2 - We describe our continuing efforts to improve the UW-SRI-ICSI Mandarin broadcast speech recognizer. This includes increasing acoustic and text training data, adding discriminative features, incorporating frame-level discriminative training criterion, multiple-pass acoustic model (AM) cross adaptation, language model (LM) genre adaptation and system combination. The net effect without LM adaptation was a 24%-64% relative reduction in character error rates (CERs) on a variety of test sets. In addition, LM adaptation gave us another 6% of relative CER reduction on broadcast conversations.

AB - We describe our continuing efforts to improve the UW-SRI-ICSI Mandarin broadcast speech recognizer. This includes increasing acoustic and text training data, adding discriminative features, incorporating frame-level discriminative training criterion, multiple-pass acoustic model (AM) cross adaptation, language model (LM) genre adaptation and system combination. The net effect without LM adaptation was a 24%-64% relative reduction in character error rates (CERs) on a variety of test sets. In addition, LM adaptation gave us another 6% of relative CER reduction on broadcast conversations.

KW - Character error rates

KW - Discriminative features

KW - Discriminative training

KW - LM adaptation

KW - Mandarin

UR - http://www.scopus.com/inward/record.url?scp=56149122078&partnerID=8YFLogxK

M3 - Conference article published in proceeding or book

SN - 9781605603162

VL - 4

SP - 2876

EP - 2879

BT - International Speech Communication Association - 8th Annual Conference of the International Speech Communication Association, Interspeech 2007

T2 - 8th Annual Conference of the International Speech Communication Association, Interspeech 2007

Y2 - 27 August 2007 through 31 August 2007

ER -

Advances in mandarin broadcast speech recognition

Abstract

Conference

Keywords

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this