From source to brain : modelling sound propagation and localisation in rooms

Sheaffer, J

Abstract

Human localisation of sound in enclosed spaces is a cross-disciplinary research topic, with important applications in auditory science, room acoustics, spatial audio and telecommunications. By combining an accelerated model of $3$D sound propagation in rooms with a perceptual model of spatial processing, this thesis provides an integrated framework for studying sound localisation in enclosed spaces on the horizontal plane, with particular emphasis on room acoustics applications. The room model is based on the finite difference time domain (FDTD) method, which has been extended to include physically-constrained sources and binaural receivers based on laser-scanned listener geometries. The underlying algorithms have been optimised to run on parallel graphics hardware, thus allowing for a high spatial resolution, and accordingly, a significant decrease of numerical dispersion evident in the FDTD method. The perceptual stage of the model features a signal processing chain emulating the physiology of the auditory periphery, binaural cue selection based on interaural coherence, and a final decision maker based on supervised learning. The entire model is shown to be capable of imitating human sound localisation in different listening situations, including free field conditions and at the presence of sound occlusion, diffraction and reflection. Results are validated against subjective data found in the literature, and the model's applications to the fields of room acoustics and spatial audio are demonstrated and discussed.

Citation

Sheaffer, J. From source to brain : modelling sound propagation and localisation in rooms. (Thesis). University of Salford

• Thesis Type: Thesis
• Deposit Date: Jul 5, 2013
• Publicly Available Date: Jul 5, 2013
• Keywords: fdtd acoustics modeling simulation psychoacoustics sound localization
• Award Date: May 15, 2013