An Efficient Assessment Methodology for the Diagnosis of Neurological Disorders from the Uttered Speech Signal

Automatic regular monitoring of the people with speech dysfunctions and the identification of the diseases has a considerable effect in the civilization. Recent days, researches are going on to find the correlation between the speech dysfunctions and the speech impairments. Many algorithms are there to forecast the sternness of the dysfunction's signs. The dysfunctions are majorly classified into Dysarthria & Apraxia. The goal of this paper is to diagnose the people who are affected by the Dysarthria and Apraxia, the diseases under the category taken into account here are Parkinson's disease, Cerebral Palsy, Aphasia and Throat tumor. The total survey on the voice signals of the subjects was made and the dysponia attributes namely Jitter, Shimmer, Pitch, Entropy, Glottal to Noise Equivalent Ratio, Harmonics to Noise Ratio and Wavelet based Dynamic Mel Frequency Coefficients have been extracted from the recorded speech signals using their respective algorithms. Besides this, the categorization has been done using the Eigen value based Support Vector Machine (SVM) classifier to sort out the subjects. The proposed algorithm attained 97.83% accuracy. We visualize our work as the improvement headed for the advancement of acoustic signal processing for the automated unremitting scrutinizes.