A Study on the Search of the Most Discriminative Speech Features in the Speaker Dependent Speech Emotion Recognition

摘要

Expressing emotion to others and recognizing emotion state of the counterpart are not difficult for human. Emotion state of a person may be recognized from the facial expression, voice, and/or gesture. Speech emotion recognition research gained a lot of attention in recent years. One of the important subjects in speech emotion recognition research is the feature selection. The speech features used will greatly influence the recognition rate. In this research, we try to find the most discriminative features for emotion recognition out from a set of 78 features. We use these features to study the feature characteristics for individual speaker by using a GMM classifier. We obtained an average of 71% recognition rate in speaker dependent case while an average of 48% recognition rate in speaker independent case.