Emotion recognition on speech signals using machine learning

With the increase in man to machine interaction, speech analysis has become an integral part in reducing the gap between physical and digital world. An important subfield within this domain is the recognition of emotion in speech signals, which was traditionally studied in linguistics and psychology. Speech emotion recognition is a field having diverse applications. The prime objective of this paper is to recognize emotions in speech and classify them in 7 emotion output classes namely, anger, boredom, disgust, anxiety, happiness, sadness and neutral. The proposed approach is based upon the Mel Frequency Cepstral coefficients (MFCC) and energy of the speech signals as feature inputs and uses Berlin database of emotional speech. The features extracted from speech are converted into a feature vector, which in turn are used to train different classification algorithms namely, Support Vector Machine (SVM), Random Decision Forest and Gradient Boosting. Random forest was found to have the highest accuracy and predicted correct emotion 81.05% of the time.