EMOTION DETECTION SYSTEM FOR HAPPY, ANGRY, NEUTRAL, FEAR, SURPRIZE AND SAD EMOTION USING PITCH, VOICED/UNVOICED ENERGY RATIO, JITTER AND DEVIATION OF PITCH

摘要

The system is designed for detecting the emotions like happy, angry, sad, neutral, surprize and fear from the speech signal. The system operates in two modes, training and detection mode. It is a multistage multiclass system. The system uses emotion specific parameters. A single parameter is in a position to classify the 6 motions in 2 categories. Happy, angry and surprize on one side and neutral, sad and fear on the other side. The classification of the three emotions in each class is carried out using again a single parameter. The system is computationally effective and gives more than 90% accuracy for all emotions.