Audio Classification in Speech and Music by using Neural Network: Multilayer Perceptron

Abstract

Audio classification can be used in many different applications. Rapid increase in the amount of audio data demands for an efficient method to automatically segment or classify audio stream based on its content. This paper focuses the attention on audio classification in music and speech. This audio classification system consists of three processing stages: feature extraction, training and classification. Spectral flux, short time energy and cepstrum coefficient are used to classify input audio into two types: speech and music. In this paper, single type of audio classification system is based on Multilayer Perceptron (MLP) neural network. Back propagation algorithm is used to perform training process. Simulation results are also included in this paper. It can classify audio files combining speech and music.