5th International Conference on Spoken Language Processing

Sydney, Australia
November 30 - December 4, 1998

Neural Network Based Pronunciation Modeling With Applications To Speech Recognition

Toshiaki Fukada (1), Takayoshi Yoshimura (2), Yoshinori Sagisaka (1)

(1) ATR-ITL, Japan
(2) Nagoya Institute of Technology, Japan

We propose a method for automatically generating a pronunciation dictionary based on a pronunciation neural network that can predict plausible pronunciations (realized pronunciations) from canonical pronunciations. This method can generate multiple forms of realized pronunciations using the pronunciation network. Experimental results on spontaneous speech show that the automatically-derived pronunciation dictionary gives consistently higher recognition rates than a conventional dictionary.

Full Paper

Bibliographic reference.  Fukada, Toshiaki / Yoshimura, Takayoshi / Sagisaka, Yoshinori (1998): "Neural network based pronunciation modeling with applications to speech recognition", In ICSLP-1998, paper 0658.