The Reality of Virtual Voice Production in Performance Spaces

Virtual reality opens the door to creating experiences that can take the user wherever or whenever they want, potentially even to places that are inaccessible to many people. Over the last decade, there has been a sharp increase in the popularity and availability of commercial VR headsets for both the worlds of entertainment and research. This has led to a revived interest in improving audio immersion in virtual environments, as a way to establish greater immersion overall.

This thesis provides an overview into various methods of auralisation for virtual spaces, investigating our ability to firstly recreate real acoustic environments naturally, and secondly to discover how true these models need to be to their corresponding real environment to maintain immersion, and what specific cues contribute most to a person's ability to feel fully immersed. Results show that for spaces in which a person is only listening to audio cues, an approximation of a real aural environment is sufficient to immerse a participant, showing a slight tendency of preference towards the falsified model in the case where a user was free to move about in a reverberant acoustic environment. Furthermore, when presented with the opportunity to freely manipulate reverberant tail length and first reflection delay in real-time for their own singing voices, immersion was maintained for users even when results differed slightly from the real measured values.