Emotion Recognition through Speech Signal for Human-Computer Interaction

摘要

This paper aims at developing a Speaker Emotion Recognition (SER) system to recognize seven different emotions namely anger, boredom, fear, disgust, happiness, neutral and sadness with a generalized feature set in real-time. Continuous HMM and LIBSVM classifiers are considered in this paper. The choice of LIBSVM classifier provides better recognition rates for few emotions (Anger and Fear) compared to the Continuous HMM classifier used in the earlier work by Xiang Li. The Hilbert-Huang transform (HHT) and Teager Energy Operator (TEO) based features gives the advantage of self-adaptability and hence can be used for real time applications.