Paper Title
SVM and Hmm Classifiers Based Speech Emotion Recognition Using Spectral Features

Abstract
In this paper we have develops an approach to speech based emotion recognition based on spectral features and combination of SVM and HMM classifier. The proposed system is divided in to two parts- audio feature extraction and verification of emotion. In the feature extraction part, first we are selecting audio clip which has to be tested, then we are extracting four spectral features from the input audio clip namely centroid, spread, flatness and projection. We are training the model before testing. For training purpose we are generating audio database which consists of speech utterances belongs to ten different emotions. After feature extraction, in the stage of emotion verification we are verifying emotion by using our classifier SVM and HMM. Index Terms- Speech Emotion Recognition, Support Vector Machines, Hidden Markov Models, Spectral Features