HERS:Human emotion recognition system

HERS is a real time system for emotion recognition through the analysis of facial expression and speech features. The system simultaneously detects emotion from live video audio streams. The system splits the video stream into sequence of images, then automatically detects the face of a person from the frames and extracts his/her emotion. The system also splits the audio stream into frames, extract the Mel-Frequency Cepstral Coefficients(MFCC) features and extract his/her emotion. This system is the prototype of an intelligent system which uses Support Vector Machines (SVM) for guided classification of emotions. Since it is a bimodal system, a higher recognition rate is expected.

I. Introduction

Emotions playa key role in human interactions. In order to create an Artificial Intelligence(AI) capable of rich machine human interaction, the AI needs to learn to recognize and respond according to human emotional states. Weare taking a step in developing such a system capable of recognizing the real time emotion of a person. It is a bimodal system that analyzes the facial expressions and speech features of a person to predict his/her emotion in real time.