Speeh/Music classification by using statistical neural networks

Abstract

This paper represents a framework for speech/music classification by using statistical neural networks. Zero Crossing Rate, Root Mean Square Power and Spectral Centroid were used as features. A dataset include 150 audio instances were labeled manually and 105 of them were used to train different networks which are PNN, GRNN and RBF. The remaining of the dataset were used as test item. Training and test performances of these three network types were discussed.