Computer-aided language learning tries to have computers serve as virtual language tutors to help people in learning non-native languages in the globalized world nowadays. In this paper we propose a framework to incorporate specially designed discriminative models with carefully trained generative models for the task of pronunciation error pattern detection. For each phoneme we train one or more SVMs with varying targets and different weights to integrate with HMM/GMMs for optimizing the detection performance from different aspects. Experiments show this integration framework effectively enhance mispronunciation detection performance.

Index Terms: Error Pattern, Viterbi Decoding, Discriminative Score, SVM