Fourth European Conference on Speech Communication and Technology

Madrid, Spain
September 18-21, 1995

Non-Uniform Unit HMMS for Speech Recognition

Takeshi Matsumura (1), Shoichi Matsunaga (2)

(1) Mitsubishi Electric Corporation, Computer & Information, Systems Laboratory, Kamakura, Kanagawa, Japan
(2) ATR Interpreting Telecommunications, Research Laboratories, Soraku-gun, Kyoto, Japan

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 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 unit model is confirmed by a phrase recognition test using an LR parser. Recognition results show that non-uniform unit HMMs achieve higher performance than conventional phoneme-unit HMMs and suggest the potential capacity of non-uniform unit HMMs.

Full Paper

Bibliographic reference.  Matsumura, Takeshi / Matsunaga, Shoichi (1995): "Non-uniform unit HMMS for speech recognition", In EUROSPEECH-1995, 499-502.