Emotion recognition in speech using MFCC and wavelet features

Recognition of emotions from speech is one of the most important sub domains in the field of affective computing. Six basic emotional states are considered for classification of emotions from speech in this work. In this work, features are extracted from audio characteristics of emotional speech by Mel-frequency Cepstral Coefficient (MFCC), and Subband based Cepstral Parameter (SBC) method. Further these features are classified using Gaussian Mixture Model (GMM). SAVEE audio database is used in this work for testing of Emotions. In the experimental results, SBC method out performs with 70% in recognition compared to 51% of recognition in MFCC algorithm.