The SVM based on SMO Optimization for Speech Emotion Recognition

With the development of artificial intelligence,speech emotion recognition (SER) increasingly has been integrated into our lives, such as medical and education. It has always been a subject that researchers have continuously explored and challenged for enhancing the robustness and generalization of the emotion recognition model. In the research process, a good classification algorithm is one of the crucial links in the entire emotion recognition process. The optimization algorithm for classification will greatly improve the recognition rate of the model. This paper proposes an improvement in the classification that applies the nonlinear support vector machine based on the optimization algorithm SMO. We enhance the recognition rate of the emotion with the help of the proposed algorithm. In order to prove its effectiveness, we do the experiment with speaker-dependent. speaker-independent and cross-corpus respectively. And the experiment results show that the proposal achieves 78.88% recognition accuracy on average.