Speech Emotion Recognition using Mel Frequency Cepstral Coefficient and SVM Classifier

Speech Emotion Recognition is the forefront of Human Machine Interaction. In this paper we present a supervised learning approach for Speech Emotion Recognition (SER) using Mel Frequency Cepstral Coefficient (MFCC) feature extraction. Using dimensionality reduction and normalization we increase the efficiency of prediction accuracy. The developed model was trained and tested against the Berlin Database of Emotional Speech recorded by the Technical University Berlin. The model was classified using a Support Vector Machine (SVM). Using validation, the model resulted with highest accuracy of 92.16% against the test samples.