iRepository at Perpustakaan UniMAP >
UNIVERSITY LIBRARY >
Conference Papers >
Please use this identifier to cite or link to this item:
|Title: ||Discrimination of pathological voices using systole activated neural network|
|Authors: ||Murugesa Pandiyan, Paulraj, Prof. Madya Dr,|
Sazali, Yaacob, Prof. Dr.
Hariharan, Muthusamy, Dr.
|Keywords: ||Acoustic features;Neural network;Back propagation algorithm;Systole activation function|
|Issue Date: ||27-Nov-2007|
|Publisher: ||Noise, Vibration and Comfort Research Group|
|Citation: ||p. 160-165|
|Series/Report no.: ||Proceedings of the Regional Conference on Engineering Mathematics, Mechanics, Manufacturing & Architecture (EM3 ARC) 2007|
|Abstract: ||The discrimination of normal and pathological voices using noninvasive acoustical analysis features helps speech specialits to perform accurate diagnoses of vocal and voices disease. Acoustic analysis is a non-invasive technique based on digital processing of the speech, acoustic analyses of normal and pathological voices have become increasingly interesting to researchers in ENT and speech pathologies. This paper presents discrimination of pathological voices using Artificial Neural Network for the accurate diagnosis of vocal and voices disease. A Neural network is trained using Back propagation algorithm with bipolar activation function and systole activation function. The neural network trained by using back propagation algorithm with systole activation function provides very promising classification accuracy of 99% to discriminate the voices as pathological or a non-pathological or a non-pathological voice accurately.|
|Description: ||Proceedings of the Regional Conference on Engineering Mathematics, Mechanics, Manufacturing & Architecture (EM3 ARC) 2007 was jointly organized by Universiti Kebangsaan Malaysia (UKM), 27th - 28th November 2007 at Kuala Lumpur, Malaysia.|
|Appears in Collections:||Conference Papers|
Sazali Yaacob, Prof. Dr.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.