Voice Pathology Identification using Deep Neural Networks

Abstract

The human voice construction is a complex biological mechanism capable of Changing pitch and volume. Some Internal or External factors frequently damage the vocal cords and change quality of voice or do some alteration in the voice modulation. The effects are reflected in expression of speech and understanding of information said by the person. So it is important to examine problem at early stages of voice change and overcome from this problem. ML play a major role in identifying whether voice is pathological or normal in nature. Voice features are extracted by Implementing Mel-frequency Cepstral Coefficients (MFCC) method, and examined on the Convolutional Neural Network (CNN) to identify the category of voice