Real Time Speech Emotion Recognition System

Speech is the important fundamental way of expressing different emotions by a person. Various developments has recently received in the field of speech emotion recognition. The objective of this paper is to perform a real-time recognition of emotion from speech, by using the Decision Tree, Support Vector Machine (SVM) and Random Forests as classifiers for classifying five different emotions. The real time recognition system is achieved with the help of a Graphical User Interface (GUI). The feature set used includes statistics of pitch, intensity and Mel Frequency Cepstral Coefficients (MFCCs). The proposed method has been verified using two self made emotional speech databases of males coming under different age groups and also with a standard database, eNTERFACE. The performance analysis in emotion recognition was done using confusion matrices, and a peak accuracy of 99.1% was yielded by random forests classifiers. And the average time taken by the GUI for emotion recognition was 1.02 sec.