An effective automatic speech emotion recognition for Tamil language based on DWT and MFCC using Stability-plasticity dilemma Neural network

In this paper we put effort on Effective Automatic Speech emotion recognition on Human computer Interaction for Tamil speech. Due to the unavailability of Tamil language database for emotion recognition, we built a database of emotional speech in Tamil. This database consists of 19 wave clips modulated with anger, joy, fear, neutral, and sad. Then we extract cepstral based features like MFCC and Discrete Wavelet Transform (DWT). A German Corpus — Berlin Database of Emotional Speech and our self built Tamil emotional databases are used for training purpose. The Neural network is trained using Stability-plasticity dilemma. Finally results are compared and explained.