Speaker Independent Emotion Recognition System - SIERS

Most speech emotion recognition system proposed to date uses the Hidden Markov Model (HMM). Such system is compute intensive and would require longer training and testing time that may not be suitable for online or smart phone application. In this paper the authors propose an alternative approach to realize the speech emotion recognition system by using Short Time Mel Frequency Cepstral Coefficient (ST-MFCC) as the features extraction method. The Speaker Independent Emotion Recognition System (SIERS) performance is measured based on three neural network and fuzzy neural network architecture; namely: Multi Layer Perceptron (MLP), Adaptive Neuro Fuzzy Inference System (ANFIS) and Generic Self-organising Fuzzy Neural Network (GenSoFNN).