An automatic pronunciation teaching system for Chinese to learn English

This paper presents the structure of a computer assisted pronunciation teaching system. The system is intended to help the English learners whose mother tongues are Chinese improve their pronunciation level of English. It is mainly based on the technology of continuous speech recognition and the research of English-Chinese comparative phonetics. The collection and labeling method of the speech data corpus are described in the paper. The algorithms of scoring and the results of some experiments are given. The scoring includes likelihood estimation based on hidden Markov model (HMM), duration score, and stress judgment. The experiment results show the system´s valuable capability on automatic pronunciation teaching.