Speech emotion recognition based on Gaussian kernel nonlinear proximal support vector machine

For the sake of improving the precision of speech emotion recognition, this paper proposed a novel speech emotion recognition approach based on Gaussian Kernel Nonlinear Proximal Support Vector Machine (PSVM) to recognize four basic human emotions (angry, joy, sadness, surprise). Firstly, preprocess speech signal containing sampling, quantification, pre-emphasizing, framing, adding window and endpoint detection. Secondly, extract speech prosody features and speech quality features. Finally, use six Nonlinear Proximal support vector machines to recognize the emotion, and then obtain the final recognition result through majority voting principle. To evaluate the approach, we also use Support vector machine (SVM) to recognize the emotion. The experimental results show that the novel approach is effective and high-speed for emotion recognition.