Background
Speech is the most complex, rapid function our body performs. In an instant, our brain decides what to say and how to say it and then precisely coordinates over 100 muscles to produce around four syllables per second. There is growing evidence that the rhythm and acoustics of someone’s voice, as well as what they are saying, reveal essential things about their mental and physical health.
Novelty and Importance
This project aims to leverage a responsible AI framework combining speech processing, advanced statistical techniques and artificial intelligence (AI) to provide a comprehensive picture of health-related changes in speech. The applicant will have access to a variety and depth of longitudinal speech data from clinical cohorts that are unique in the field. The findings have the potential to revolutionise our understanding of speech patterns, paving the way for innovative applications in healthcare assessments, personalised therapy plans, cognitive assessment, and beyond.
Aims and Objectives
To develop an analytical framework to characterise and track changes in key symptoms of psychological and neurological disorders using speech markers. The student will learn the signal processing and machine learning techniques required to extract relevant features from RMT-collected speech signals, then assess the suitability of these features for assessing longitudinal changes in speech. This parameterisation will include the incorporation of both knowledge-driven and data-driven approaches, as well as the use of speech and language foundational models. Finally, they will use advanced analytical methods to determine critical temporal dynamics and the evolution of different speech properties in each clinical population.

