Development of an intelligent headset system for continuous speech monitoring and feedback in Parkinson’s Disease: bridging clinical therapy and daily life
Author(s)
Benichou, Yacine
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
This thesis concentrates on the creation and assessment of a thorough environmental monitoring system in the form of a headset targeted at enhancing speech intelligibility and classifying ambient sounds in ICU environments, particularly focusing on Parkinson’s Disease (PD). The system is intended to continually evaluate speech quality, observe interactions between patients and staff, and categorize ambient sounds in real-time to improve patient care and comfort. An essential element of the system is the implementation of advanced speech intelligibility metrics such as the Speech Intelligibility Index (SII) and Speech Transmission Index (STI), which deliver quantitative evaluations of speech clarity across different acoustic settings. However, because traditional techniques face challenges in managing speech distortions and noise, especially in healthcare contexts, a machine learning strategy was investigated to classify sounds and measure speech intelligibility more efficiently.
The system utilizes real-time information from multiple sensors, including microphones, to analyze and categorize audio signals based on characteristics like articulation rate and speech signal variations. The thesis encompasses a comprehensive evaluation of hypokinetic dysarthria, a speech impairment prevalent in PD patients, which notably affects articulation and speech intelligibility. By utilizing feature extraction methods and classification algorithms, the system is capable of autonomously identifying and distinguishing between normal and impaired speech, offering insights into communication difficulties associated with PD. The findings of this research indicate that a machine learning-driven method presents a hopeful avenue for real-time observation and feedback within clinical settings, enabling more precise and adaptable speech evaluations.
This research also considers the wider impacts of sound and speech monitoring systems within healthcare environments, where they can enhance interactions between patients and staff while minimizing noise pollution, ultimately leading to improved patient results. Furthermore, the study emphasizes the possible applications of this system within clinical therapy for PD patients, providing ongoing feedback to support enhancements in speech clarity and articulation precision outside of therapy. The thesis wraps up with suggestions for additional research into practical applications of this technology and its prospective incorporation into long-term healthcare monitoring systems.
Sponsor
Date
2024-12-09
Extent
Resource Type
Text
Resource Subtype
Thesis