Cantonese Speech Recognition System
(L. W. Chan)

Cantonese is a monosyllabic and tonal language which the tonal information plays a very significant role in the recognition. This project is to design and construct a speaker independent speech recognition system for Cantonese which contains more than 1000 syllables. We use the recurrent neural network model and the modular neural network approach in our system. The recognition of tones, isolated words and continuous speech are under investigation.