Automatic Pronunciation Evaluation Based on Feature Extraction and Combination

This paper presents an effective method for automatic pronunciation evaluation, which is based on feature extraction and combination. The proposed system extracts different kinds of evaluation features and combines them to produce an ultimate machine score, which predicts the overall pronunciation quality of a student. Experiments on a reading speech database show that most of the selected features are distinctive features for pronunciation quality, which have strong correlations with human scores. In addition, the combination of different features using linear regression (IR) can achieve better performance than using individual features and the produced machine scores are comparable to human scores.