Voice Pathology Classification System Using Machine Learning

Vocal disorders are pathological states that discomfort the quality of speech which is produced by the voice box or larynx. The disrupted nature of voice causes inflammation to larynx or voice box mainly due to overuse or irritation or infection. The goal is to build a machine learning model which categorizes distinct class of diseased condition of vocal chords, include Dysphonia or spasmodic dysphonia, Normal, Stammering which is an instance of stuttering and Vocal palsy or vocal fold paralysis from AIISH (All India Institute of Speech and Hearing), voice data repository and voice from individuals. The machine learning classifiers used to handle the classification problem of vocal disorders are Support Vector Machine and K-Nearest Neighbor. The resulted outcome is evaluated based on the features of voice selected from the process of feature extraction using MFCC (Mel Frequency cepstral coefficients).