Large vocabulary Taiwanese (Min-nan) speech recognition using tone features and statistical pronunciation modeling

Dau Cheng Lyu; Min Siong Liang; Yuang Chin Chiang; Chun Nan Hsu; Ren Yuan Lyu

Large vocabulary Taiwanese (Min-nan) speech recognition using tone features and statistical pronunciation modeling

Dau Cheng Lyu, Min Siong Liang, Yuang Chin Chiang, Chun Nan Hsu, Ren Yuan Lyu

Department of Computer Science and Information Engineering

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

14 Scopus citations

Abstract

A large vocabulary Taiwanese (Min-nan) speech recognition system is described in this paper. Due to the severe multiple pronunciation phenomenon in Taiwanese partly caused by tone sandhi, a statistical pronunciation modeling technique based on tonal features is used. This system is speaker independent. It was trained by a bi-lingual Mandarin/Taiwanese speech corpus to alleviate the lack of pure Taiwanese speech corpus. The searching network is constructed based on nodes of Chinese characters and results in the direct output Chinese character string. Experiments show that by using the approaches proposed in this paper, the character error rate can decrease significantly from 21.50% to 11.97%.

Original language	English
Pages	1861-1864
Number of pages	4
State	Published - 2003
Event	8th European Conference on Speech Communication and Technology, EUROSPEECH 2003 - Geneva, Switzerland Duration: 01 09 2003 → 04 09 2003

Conference

Conference	8th European Conference on Speech Communication and Technology, EUROSPEECH 2003
Country/Territory	Switzerland
City	Geneva
Period	01/09/03 → 04/09/03

Cite this

@conference{101c7c04758b4db4bd78514972c9b234,

title = "Large vocabulary Taiwanese (Min-nan) speech recognition using tone features and statistical pronunciation modeling",

abstract = "A large vocabulary Taiwanese (Min-nan) speech recognition system is described in this paper. Due to the severe multiple pronunciation phenomenon in Taiwanese partly caused by tone sandhi, a statistical pronunciation modeling technique based on tonal features is used. This system is speaker independent. It was trained by a bi-lingual Mandarin/Taiwanese speech corpus to alleviate the lack of pure Taiwanese speech corpus. The searching network is constructed based on nodes of Chinese characters and results in the direct output Chinese character string. Experiments show that by using the approaches proposed in this paper, the character error rate can decrease significantly from 21.50% to 11.97%.",

author = "Lyu, {Dau Cheng} and Liang, {Min Siong} and Chiang, {Yuang Chin} and Hsu, {Chun Nan} and Lyu, {Ren Yuan}",

year = "2003",

language = "英语",

pages = "1861--1864",

note = "8th European Conference on Speech Communication and Technology, EUROSPEECH 2003 ; Conference date: 01-09-2003 Through 04-09-2003",

}

Large vocabulary Taiwanese (Min-nan) speech recognition using tone features and statistical pronunciation modeling. / Lyu, Dau Cheng; Liang, Min Siong; Chiang, Yuang Chin et al.
2003. 1861-1864 Paper presented at 8th European Conference on Speech Communication and Technology, EUROSPEECH 2003, Geneva, Switzerland.

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

TY - CONF

T1 - Large vocabulary Taiwanese (Min-nan) speech recognition using tone features and statistical pronunciation modeling

AU - Lyu, Dau Cheng

AU - Liang, Min Siong

AU - Chiang, Yuang Chin

AU - Hsu, Chun Nan

AU - Lyu, Ren Yuan

PY - 2003

Y1 - 2003

N2 - A large vocabulary Taiwanese (Min-nan) speech recognition system is described in this paper. Due to the severe multiple pronunciation phenomenon in Taiwanese partly caused by tone sandhi, a statistical pronunciation modeling technique based on tonal features is used. This system is speaker independent. It was trained by a bi-lingual Mandarin/Taiwanese speech corpus to alleviate the lack of pure Taiwanese speech corpus. The searching network is constructed based on nodes of Chinese characters and results in the direct output Chinese character string. Experiments show that by using the approaches proposed in this paper, the character error rate can decrease significantly from 21.50% to 11.97%.

AB - A large vocabulary Taiwanese (Min-nan) speech recognition system is described in this paper. Due to the severe multiple pronunciation phenomenon in Taiwanese partly caused by tone sandhi, a statistical pronunciation modeling technique based on tonal features is used. This system is speaker independent. It was trained by a bi-lingual Mandarin/Taiwanese speech corpus to alleviate the lack of pure Taiwanese speech corpus. The searching network is constructed based on nodes of Chinese characters and results in the direct output Chinese character string. Experiments show that by using the approaches proposed in this paper, the character error rate can decrease significantly from 21.50% to 11.97%.

UR - http://www.scopus.com/inward/record.url?scp=84961715459&partnerID=8YFLogxK

M3 - 论文

AN - SCOPUS:84961715459

SP - 1861

EP - 1864

T2 - 8th European Conference on Speech Communication and Technology, EUROSPEECH 2003

Y2 - 1 September 2003 through 4 September 2003

ER -

Large vocabulary Taiwanese (Min-nan) speech recognition using tone features and statistical pronunciation modeling

Abstract

Conference

Other files and links

Fingerprint

Cite this