HMM continuous speech recognition using stochastic language models

Author

Kita, Kenji ; Kawabaa, T. ; Hanazawa, Toshiyuki

Author_Institution

ATR Interpreting Telephony Res. Lab., Kyoto, Japan

fYear

1990

fDate

3-6 Apr 1990

Firstpage

581

Abstract

Three stochastic language models are investigated for hidden Markov model (HMM) continuous-speech recognition system. They are the trigram model of Japanese syllables, the stochastic shift/reduce model in LR parsing, and the trigram model of context-free rewriting rules. These stochastic language models are incorporated into the HMM-LR continuous-speech recognition system. The phrase recognition rate is improved from 72.4% to 81.0%. Moreover, for a high-quality HMM-LR speech recognition system which uses separate vector quantization (VQ) and fuzzy VQ, the phrase recognition rate is improved from 88.2% to 93.2%, and a rate of 100% is achieved for the top four choices

Keywords

Markov processes; context-free grammars; natural languages; speech recognition; Japanese syllables; LR parsing; context-free rewriting rules; continuous speech recognition; fuzzy VQ; hidden Markov model; stochastic language models; stochastic shift/reduce model; trigram model; vector quantization; Cepstral analysis; Context modeling; Fuzzy systems; Hidden Markov models; Natural languages; Predictive models; Probability; Speech recognition; Stochastic processes; Stochastic systems; Vector quantization;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 1990. ICASSP-90., 1990 International Conference on

Conference_Location

Albuquerque, NM

ISSN

1520-6149

Type

conf

DOI

10.1109/ICASSP.1990.115779

Filename

115779