A REVIEW PAPER ON EMOTION RECOGNITION

In the past decade a lot of research has gone into Speech Emotion Recognition (SER). Recognition of emotion is always a difficult problem, particularly if the recognition of emotion is done by using speech signal. In human machine interface application, emotion recognition from the speech signal has been research topic since many years. This paper reviews similar works done in the domain of Speech Emotion Recognition (SER) system. It highlights various feature extraction methods, prosodic and spectral features, and various models, Gaussian Mixture Model (GMM), k-Nearest Neighbours (k-NN), Support Vector Machine (SVM), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN) etc, used resulting in varying levels of accuracy. After analysing various similar papers, this paper attempts to provide a comparative study on effect of models and feature vectors on SER system accuracy. Keywords— Speech Emotion Recognition (SER), Prosodic Feature, Spectral Feature, GMM, k-NN, SVM, CNN, RNN