PhD Studentship in Modelling Dialogue Understandability in Media Using AI and Listener Types

This fully funded PhD studentship at the University of Salford focuses on advancing the understanding of dialogue comprehension in media. Led by Dr. Ben Shirley and Dr. Rebecca Vos, in collaboration with industry partners at Xperi, the project aims to create a listener-aware framework for quantifying dialogue understandability. By integrating AI, speech models, and large language models, the research will explore how listener profiles, listening environments, and cognitive effort impact comprehension. The goal is to develop a validated computational model that improves accessibility and quality in streaming, broadcast, and media production.

• Develop a listener-aware framework for predicting dialogue comprehension across diverse audiences and listening environments.
• Design and implement quantitative metrics for measuring dialogue understandability beyond traditional speech intelligibility.
• Model listener types, including hearing profiles, language proficiency, and accent familiarity.
• Explore the use of AI, speech models, and large language models to predict comprehension and listening effort.
• Develop validated computational models and evaluation protocols for media content.
• Conduct experiments using existing datasets and newly collected listener data to analyze listening contexts and cognitive effort.
• Publish findings in academic journals and present results at relevant conferences.