Use of prosodic information for Mandarin word post-recognition

Chung-Hsien Wu, Yeou Jiunn Chen

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

Abstract

In this paper, a two-stage recognition scheme, phonetic recognition followed by prosodic recognition, is established. In the phonetic recognition process, 21 INITIAL and 37 FINAL context-independent HMM's are used to construct the phonetic recognizer. In the prosodic recognizer, 175 context-dependent prosodic HMM's are used to model the complicated tone behavior for all possible tone concatenations. Five anti-prosodic HMM's, each corresponding to one lexical tone, are constructed to enhance the discriminability among prosodic HMM's. This system was evaluated in a speaker-dependent mode on a vocabulary size of thirty thousand words. The experimental results show that the recognition rate was improved from 80.3% to 86.7% using the prosodic information.

Original languageEnglish
Title of host publicationSpeech and Image Technologies for Computing and Telecommunications
EditorsM. Deriche, M. Moody, M. Bennamoun
PublisherIEEE
Pages253-256
Number of pages4
Volume1
Publication statusPublished - 1997
EventProceedings of the 1997 IEEE TENCON Conference. Part 1 (of 2) - Brisbane, Australia
Duration: 1997 Dec 21997 Dec 4

Other

OtherProceedings of the 1997 IEEE TENCON Conference. Part 1 (of 2)
CityBrisbane, Australia
Period97-12-0297-12-04

Fingerprint

Speech analysis

All Science Journal Classification (ASJC) codes

  • Engineering(all)

Cite this

Wu, C-H., & Chen, Y. J. (1997). Use of prosodic information for Mandarin word post-recognition. In M. Deriche, M. Moody, & M. Bennamoun (Eds.), Speech and Image Technologies for Computing and Telecommunications (Vol. 1, pp. 253-256). IEEE.
Wu, Chung-Hsien ; Chen, Yeou Jiunn. / Use of prosodic information for Mandarin word post-recognition. Speech and Image Technologies for Computing and Telecommunications. editor / M. Deriche ; M. Moody ; M. Bennamoun. Vol. 1 IEEE, 1997. pp. 253-256
@inproceedings{14b8e75f5d6e48b1883b7ba597109289,
title = "Use of prosodic information for Mandarin word post-recognition",
abstract = "In this paper, a two-stage recognition scheme, phonetic recognition followed by prosodic recognition, is established. In the phonetic recognition process, 21 INITIAL and 37 FINAL context-independent HMM's are used to construct the phonetic recognizer. In the prosodic recognizer, 175 context-dependent prosodic HMM's are used to model the complicated tone behavior for all possible tone concatenations. Five anti-prosodic HMM's, each corresponding to one lexical tone, are constructed to enhance the discriminability among prosodic HMM's. This system was evaluated in a speaker-dependent mode on a vocabulary size of thirty thousand words. The experimental results show that the recognition rate was improved from 80.3{\%} to 86.7{\%} using the prosodic information.",
author = "Chung-Hsien Wu and Chen, {Yeou Jiunn}",
year = "1997",
language = "English",
volume = "1",
pages = "253--256",
editor = "M. Deriche and M. Moody and M. Bennamoun",
booktitle = "Speech and Image Technologies for Computing and Telecommunications",
publisher = "IEEE",

}

Wu, C-H & Chen, YJ 1997, Use of prosodic information for Mandarin word post-recognition. in M Deriche, M Moody & M Bennamoun (eds), Speech and Image Technologies for Computing and Telecommunications. vol. 1, IEEE, pp. 253-256, Proceedings of the 1997 IEEE TENCON Conference. Part 1 (of 2), Brisbane, Australia, 97-12-02.

Use of prosodic information for Mandarin word post-recognition. / Wu, Chung-Hsien; Chen, Yeou Jiunn.

Speech and Image Technologies for Computing and Telecommunications. ed. / M. Deriche; M. Moody; M. Bennamoun. Vol. 1 IEEE, 1997. p. 253-256.

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

TY - GEN

T1 - Use of prosodic information for Mandarin word post-recognition

AU - Wu, Chung-Hsien

AU - Chen, Yeou Jiunn

PY - 1997

Y1 - 1997

N2 - In this paper, a two-stage recognition scheme, phonetic recognition followed by prosodic recognition, is established. In the phonetic recognition process, 21 INITIAL and 37 FINAL context-independent HMM's are used to construct the phonetic recognizer. In the prosodic recognizer, 175 context-dependent prosodic HMM's are used to model the complicated tone behavior for all possible tone concatenations. Five anti-prosodic HMM's, each corresponding to one lexical tone, are constructed to enhance the discriminability among prosodic HMM's. This system was evaluated in a speaker-dependent mode on a vocabulary size of thirty thousand words. The experimental results show that the recognition rate was improved from 80.3% to 86.7% using the prosodic information.

AB - In this paper, a two-stage recognition scheme, phonetic recognition followed by prosodic recognition, is established. In the phonetic recognition process, 21 INITIAL and 37 FINAL context-independent HMM's are used to construct the phonetic recognizer. In the prosodic recognizer, 175 context-dependent prosodic HMM's are used to model the complicated tone behavior for all possible tone concatenations. Five anti-prosodic HMM's, each corresponding to one lexical tone, are constructed to enhance the discriminability among prosodic HMM's. This system was evaluated in a speaker-dependent mode on a vocabulary size of thirty thousand words. The experimental results show that the recognition rate was improved from 80.3% to 86.7% using the prosodic information.

UR - http://www.scopus.com/inward/record.url?scp=0031381261&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0031381261&partnerID=8YFLogxK

M3 - Conference contribution

VL - 1

SP - 253

EP - 256

BT - Speech and Image Technologies for Computing and Telecommunications

A2 - Deriche, M.

A2 - Moody, M.

A2 - Bennamoun, M.

PB - IEEE

ER -

Wu C-H, Chen YJ. Use of prosodic information for Mandarin word post-recognition. In Deriche M, Moody M, Bennamoun M, editors, Speech and Image Technologies for Computing and Telecommunications. Vol. 1. IEEE. 1997. p. 253-256