Detection of pathological voices using discrete wavelet transform and artificial neural networks

The aim of this work is to develop an efficient voice disorder detection system using Discrete Wavelet Transform (DWT) and Feed Forward Neural Network (FFNN). In this experimental implementation the normal and abnormal utterances taken from Saarbrueken Voice Database (SVD) are subjected to 1-D Discrete Wavelet Decomposition and the energy of wavelet subband coefficients are computed. FFNN is finally used as a classifier to discriminate pathological voices from normal samples. The proposed system achieves 93.3% accuracy.