Training of lexical models based on DTW-based parameter reestimation algorithm

Author

Abe, Yoshiharu ; Nakajima, Kunio

Author_Institution

Inf. Syst. & Electron. Dev. Lab., Mitsubishi Electr. Corp., Kamakura, Japan

fYear

1988

fDate

11-14 Apr 1988

Firstpage

623

Abstract

A systematic method for development of word models for large vocabulary word recognition is described. The word models are comprised of successive clusters of states whose durations are governed by continuous densities. They are generated by a lexical rule and trained by an iterative algorithm based on maximum likelihood estimation and DTW temporal alignment. Several lexical rules, along with conventional DTW matching method, are experimentally evaluated by both close vocabulary and open vocabulary tests in similar word environments. The results show the effectiveness of stochastic modeling and systematic generation of word models

Keywords

parameter estimation; speech recognition; stochastic processes; DTW temporal alignment; DTW-based parameter reestimation algorithm; close vocabulary tests; continuous densities; iterative algorithm; large vocabulary word recognition; lexical models; lexical rule; maximum likelihood estimation; open vocabulary tests; speech recognition; stochastic modeling; successive state cluster; systematic generation; word models; Hidden Markov models; Information systems; Iterative algorithms; Parameter estimation; Speech recognition; Speech synthesis; Stochastic systems; Testing; Training data; Vocabulary;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 1988. ICASSP-88., 1988 International Conference on

Conference_Location

New York, NY

ISSN

1520-6149

Type

conf

DOI

10.1109/ICASSP.1988.196662

Filename

196662