Movements in Binaural Space: Issues in HRTF Interpolation and Reverberation, with applications to Computer Music

This thesis deals broadly with the topic of Binaural Audio. After reviewing the literature, a reappraisal of the minimum-phase plus linear delay model for HRTF representation and interpolation is offered. A rigorous analysis of threshold based phase unwrapping is also performed. The results and conclusions drawn from these analyses motivate the development of two novel methods for HRTF representation and interpolation. Empirical data is used directly in a Phase Truncation method. A Functional Model for phase is used in the second method based on the psychoacoustical nature of Interaural Time Differences. Both methods are validated; most significantly, both perform better than a minimum-phase method in subjective testing. The accurate, artefact-free dynamic source processing afforded by the above methods is harnessed in a binaural reverberation model, based on an early reflection image model and Feedback Delay Network diffuse field, with accurate interaural coherence. In turn, these flexible environmental processing algorithms are used in the development of a multi-channel binaural application, which allows the audition of multi-channel setups in headphones. Both source and listener are dynamic in this paradigm. A GUI is offered for intuitive use of the application. HRTF processing is thus re-evaluated and updated after a review of accepted practice. Novel solutions are presented and validated. Binaural reverberation is recognised as a crucial tool for convincing artificial spatialisation, and is developed on similar principles. Emphasis is placed on transparency of development practices, with the aim of wider dissemination and uptake of binaural technology