Towards non-uniform unit HMMs for speech recognition

Author

Matsumura, Takeshi ; Matsunaga, Shoichi

Author_Institution

ATR Interpreting Telecommun. Res. Labs., Kyoto, Japan

fYear

1994

fDate

26-27 Sep 1994

Firstpage

89

Lastpage

92

Abstract

A novel acoustic modeling algorithm that generates non-uniform unit HMMs to effectively cope with spectral variations in fluent speech is proposed. The algorithm is devised for the automatic iterative generation of long-span units for the non-uniform modeling. This generation algorithm is based on an entropy reduction criterion using text data and a maximum likelihood criterion using speech data. The effectiveness of the non-uniform models was confirmed by comparing likelihood values between the long-span unit HMMs and the conventional phoneme-unit HMMs. Preliminary results suggest that non-uniform unit HMMs achieve higher performance than phoneme-unit HMMs

Keywords

entropy; hidden Markov models; iterative methods; maximum likelihood estimation; spectral analysis; speech recognition; acoustic modeling algorithm; automatic iterative generation; entropy reduction criterion; fluent speech; long-span units; maximum likelihood criterion; nonuniform unit HMMs; performance; phoneme-unit HMMs; spectral variations; speech recognition; text data; Character generation; Concatenated codes; Context modeling; Entropy; Hidden Markov models; Iterative algorithms; Performance evaluation; Robustness; Speech recognition; Testing;

fLanguage

English

Publisher

ieee

Conference_Titel

Interactive Voice Technology for Telecommunications Applications, 1994., Second IEEE Workshop on

Conference_Location

Kyoto

Print_ISBN

0-7803-2074-3

Type

conf

DOI

10.1109/IVTTA.1994.341538

Filename

341538