An Automatic Voice Disorder Detection System Based On Extreme Gradient Boosting Classifier

Abstract

This paper describes our approach to develop an Automatic Voice Disorder Detection (AVDD) system that can diagnose the patient voice and determine if the voice is normal and healthy or disordered due to a certain pathology. Our system is based on the Saarbrucken Voice Dataset (SVD) to feed our machine-learning model, and exploiting the Mel Frequency Cepstral Coefficients (MFCC) as an extracted feature. For the classification, we chose the Extreme Gradient Boosting (XGBoost) classifier.