FEMH Voice Data Challenge: Voice disorder Detection and Classification using Acoustic Descriptors

This paper describes the participation of TCS Research and Innovation, Mumbai in the FEMH voice data challenge. The goal of the FEMH voice data challenge is detection of pathological voice and classification into three different categories using voice samples. In this work, we use a mix of speech processing and machine learning techniques to not only automatically detect pathological speech but also classify into one of the three categories namely, Neoplasm, Phonotrauma and Vocal Palsy.