Emotion recognition is a multi branched area which has attained more attention from researchers in the recent years. The main aim of utilizing emotion recognition is to improve the interface between humans and machines. The presented work introduced a new emotion recognition scheme namely 'Enhanced Gravitational Search Algorithm (EGSA) based Support Vector Neural Network (SVNN)' that is capable of recognizing the emotions in speech signals. The enhanced GSA algorithm is used in training the SVNN classifier to identify the emotions. To recognize the emotions in speech signal, the spectral features are extracted from the input signal and the extracted spectral features are given to the proposed EGSA based SVNN. Standard emotion databases such as Telugu and Berlin speech databases are utilized in simulation of the proposed EGSA based SVNN scheme. It is shown from the results that the proposed EGSA based SVNN classifier has outperformed the existing SVM classifier in terms of accuracy, FNR and FPR.
Volume 11 | 10-Special Issue
Pages: 1238-1251
DOI: 10.5373/JARDCS/V11SP10/20192969