Speech Emotion Recognition Method Based on Multi-strategy LibSVM

With the development of human–computer interaction technology, more and more attention has been paid to the research of speech emotion recognition technology, and there are many research results in the recognition of speech emotion features, but there is still a certain gap between the current research results and practical applications. Based on library for support vector machine (LibSVM), this paper proposes a multi-strategy speech emotion recognition method, which recognizes speech emotion according to the idea of stepwise classification by using the different contribution of different feature values of speech emotion to different emotion classification. Experiments show that this method is more accurate than directly using LibSVM to recognize speech emotion, which provides a new idea for speech emotion recognition research.