Automatic detection of voice disorders using self loop architecture in back propagation network

Abstract

Acoustic analysis is a non-invasive technique to detect the voice disorders and diagnose the vocal and voice disease. In the recent years, voice disease are increasing dramatically due to unhealthy social habits and voice abuse. In this paper, the detection of voice disorders based on classification of pathological voices using neural network trained by Back propagation with slope parameter improves the convergence ability of BP propagation algorithm. A simple scheme is proposed to fix the slope parameters of the bipolar sigmoidal activation function. Self loop scheme is the output of the hidden neurons feedback to itself which improved the training time and generalization of the network. The proposed algorithms provide better classification rate than conventional back propagation algorithm for the automatic detection of voices disorders.