CROSS CORPUS SPEECH EMOTION RECOGNITION

Abstract

Speech emotion recognition (SER) system plays a major role in human machine interaction. The emotion detection is natural for humans, but for machines it is tedious. Thus the proposed system aims to improve human machine interaction using emotion related information. In this paper, Mel Frequency Cepstral Coefficient (MFCC) feature is extracted from speech signal and Support Vector Machine (SVM) classifier is used to classify emotions. Performance analysis is done by considering different combinations of training and testing database and the databases considered here are Berlin, Enterface and RAVDESS.