Diagnosis of vocal fold pathology using time-domain features and systole activated neural network

Abstract

Due to the nature of job, unhealthy social habits and voice abuse, the people are subjected to the risk of voice problems. It is well known that most of vocal fold pathologies cause changes in the acoustic voice signal. Therefore, the voice signal can be a useful tool to diagnose them. Acoustic voice analysis can be used to characterize the pathological voices. This paper presents the detection of vocal fold pathology with the aid of the speech signal recorded from the patients. Time-domain features are proposed and extracted to detect the vocal fold pathology. The main advantages of this method are less computation time, possibility of real-time system development and it requires no transformation techniques (frequency transformation or time-frequency transformation). In order to test the effectiveness and reliability of the proposed time-domain features, a simple neural network model with systole activation function is proposed and trained by conventional back propagation (BP) algorithm. The classification accuracy of the proposed systole activated neural network is comparable with the results of neural network model with sigmoidal activation function. The simulation results show that the proposed systole activated neural network reduces the time taken for training the neural network.