Date of Award
6-2-2023
Publication Type
Dissertation
Degree Name
Ph.D.
Department
Electrical and Computer Engineering
Keywords
Artificial intelligence;biomedical signals;deep learning;disease diagnosis;speech;voice pathology
Supervisor
Esam Abdel-Raheem
Rights
info:eu-repo/semantics/embargoedAccess
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.
Abstract
Voice pathology is not only circumscribed by voice impairment or speech disorder. Pathological voice is also a biomarker of neuropsychiatric and neurocognitive diseases, including physical and muscular conditions. Alzheimer’s, Parkinson’s, Schizophrenia, ASD (Autism spectrum disorder), oral/lung cancer, depression, and asthma strongly correlate with voice disability. The physicians' current endoscopic procedures to detect pathological voices are painful for the patients. Clinical invasive diagnostic procedures, for example, laryngoscopy, laryngeal electromyography, stroboscopy, etc., require high-level expertise; they are expensive and time-consuming. This research focuses on establishing automated voice signal processing-based noninvasive procedures to identify pathological voices with objective diagnostics on top of subjective assessment. Several quick computerized digital signal processing-based techniques are implemented that require no extensive training/expensive equipment. Being noninvasive, they do not traumatize the patients. Also, can evaluate structural, neurological, and respiratory voice disorder. An extensive temporal, spectral, acoustical, wavelet domain based feature analysis is also performed to enhance the current understanding of pathological voice.
Recommended Citation
Islam, Rumana, "Signal Processing Based Pathological Voice Detection Techniques" (2023). Electronic Theses and Dissertations. 9350.
https://scholar.uwindsor.ca/etd/9350