Towards anti-stuttering: understanding relations between stress and stuttering

This project aims to determine the relation between stuttering and stress using ECG and GSR physiological signals. The thesis question is "Does stress or anxiety have any influence on the stuttering severity in stutterers". The physiological signals are collected from two stutterers and two non-stutterers over several reading and speaking tasks. The useful physiological markers and features are determined to compare and study the stress variations in stutterers and non-stutterers. The human autonomous nervous system branches were monitored while the subject speaks in order to study the role of changes in sympathetic and parasympathetic branches. In this study we found stuttering having inversely proportionate relationship with the HRV features. In addition, machine learning methods such as decision tree and support vector machine were investigated in this study to classify the severity level of stuttering based on these physiological features.