A new HMM/LVQ hybrid algorithm for speech recognition

Author

Katagiri, Shigeru ; Lee, Chin-Hui

Author_Institution

AT&T Bell Labs., Murray Hill, NJ, USA

fYear

1990

fDate

2-5 Dec 1990

Firstpage

1032

Abstract

It is shown that by combining the discriminative power of learning vector quantization (LVQ) training algorithms and the capability of modeling temporal variations of a hidden Markov model (HMM) into a hybrid algorithm, the performance of an HMM-based recognition algorithm is significantly improved. The hybrid algorithm was tested in a multispeaker, isolated word mode, using a highly confusable vocabulary consisting of the nine English E-set words. The average word accuracy for the original HMM-based system was 62%. When the LVQ classifier was incorporated, the word accuracy increased to 81%

Keywords

Markov processes; speech recognition; E-set words; HMM-based recognition algorithm; HMM/LVQ hybrid algorithm; LVQ classifier; LVQ training algorithm; discriminative power; hidden Markov model; highly confusable vocabulary; isolated word mode; learning vector quantization; speech recognition; temporal variations; Acoustic distortion; Artificial neural networks; Hidden Markov models; Laboratories; Pattern recognition; Speech recognition; Testing; Vector quantization; Visual perception; Vocabulary;

fLanguage

English

Publisher

ieee

Conference_Titel

Global Telecommunications Conference, 1990, and Exhibition. 'Communications: Connecting the Future', GLOBECOM '90., IEEE

Conference_Location

San Diego, CA

Print_ISBN

0-87942-632-2

Type

conf

DOI

10.1109/GLOCOM.1990.116659

Filename

116659