English sentence stress detection system based on HMM framework

Computer Assisted Language Learning based on speech technology is gaining increasing attention in recent years. It provides an interactive platform for students to learn foreign languages. It is an objective of our work to develop such a system to help students to learn English sentence stress. This paper presents a novel method based on Hidden Markov Model (HMM) to detect English sentence stress. Different features, including Mel-Frequency Cepstrum Coefficients (MFCCs), energy and pitch, are investigated to detect English sentence stress and the results are then compared. The ability of each dimension of an individual feature to differentiate stressed vowels is evaluated. The best result, 80.6%, is obtained by combining normalized pitch values and MFCCs. And our method has the advantages of speaker independent and text independent.