A study on Mandarin broadcast news speech recognition

Author

Chen, C.L. ; Wang, Y.R. ; Chen, S.H.

Author_Institution

Dept. of Commun. Eng., Nat. Chiao Tung Univ., Hsinchu, Taiwan

fYear

2004

fDate

15-18 Dec. 2004

Firstpage

257

Lastpage

260

Abstract

In this paper, a basic Mandarin broadcast news speech recognition system is constructed using the MATBN database. It considers the acoustic modeling for Mandarin base-syllables, particles, and paralinguistic phenomena. It also considers environment-dependent acoustic modeling for three recording environments: studio anchors, outdoor reporters, and outdoor interviewees. Moreover, it incorporates a bigram language model with adaptation, using the data in MATBN. Syllable recognition rates of 89.64, 84.42 and 61.62% were achieved for the three environments of anchors, reporters and interviewees, respectively.

Keywords

adaptive signal processing; hidden Markov models; linguistics; natural languages; speech recognition; HMM; Mandarin broadcast news speech recognition system; Mandarin particles; adaptive bigram language model; base-syllables; environment-dependent acoustic modeling; outdoor interviewees; outdoor reporters; paralinguistic phenomena; speech characteristics spontaneity; speech environmental variation; studio anchors; syllable recognition rates; Adaptation model; Background noise; Broadcasting; Data mining; Databases; Hidden Markov models; Signal to noise ratio; Speech enhancement; Speech recognition; XML;

fLanguage

English

Publisher

ieee

Conference_Titel

Chinese Spoken Language Processing, 2004 International Symposium on

Print_ISBN

0-7803-8678-7

Type

conf

DOI

10.1109/CHINSL.2004.1409635

Filename

1409635