Speech Emotion Recognition with Application to Mental Health A Tensor Perspective

Abstract

Speech Emotion Recognition (SER) has been an active area of research newlineever since the need for smooth and natural Human-Computer Interaction newline(HCI) came into play. This thesis aims to develop an SER system based on newlinean amalgamation of Tensor Factorization and Neural Network-based learning newlineto mitigate several issues while using contemporary deep learning architec- newlinetures. This, in turn, is helpful towards recognizing the mental health issues newlinesuch as depression, anxiety, etc., from speech signals as it is shown in the newlineliterature that mental health and emotions are highly correlated. As such, newlinethis thesis tries to provide techniques to incorporate emotional information to newlineassess mental health conditions from speech signals, thereby helping the psy- newlinechologists assign a depression score to patients based on their experience and newlinemachine-generated score, thereby mitigating any human bias which might newlinecreep in human-only situations. newline