Creation of a system for real time virtual auditory space and its application to dynamic sound localization

The study of sound localization has historically been done using stationary sources with no head movement, in part due to technological constraints. One way of studying sound localization is to use virtual auditory space (VAS) stimuli; in this approach a filtered sound is presented over headphones with the filter designed to create eardrum pressures that match those created by a loudspeaker located in real space. These filters, called spatialization filters, are designed to simulate a sound-source in an arbitrary location.; This thesis used VAS stimuli to explore how movement of the head or sound source can affect the perceived location of a sound. There are three parts to this thesis: (1) the creation of a system capable of rendering dynamic auditory scenes; (2) the development of effective spatialization filters that account for movement of the sound source relative to the head; and (3) an investigation into how subjects perform in tasks that utilize dynamic auditory stimuli. In the system design, two key aspects were the ability to change the spatialization filters dynamically and the small system delay between a head movement and the corresponding update to the filter. The key issue related to the use of Head Related Transfer Functions (HRTFs) as spatialization filters was the requirement of a smooth percept as the virtual source moves, in particular the interpolation between measured HRTFs. A set of human listening experiments were conducted to study several aspects of the perception of dynamic stimuli. Specifically, reaction times in response to dynamic sound stimuli as well as the subject's ability to follow a moving virtual source were measured.; Measurements of system performance showed that a cost effective, flexible, and open-source system could be developed that is capable of dynamic simulations with system delays less than 7 ms. In dynamic experiments, auditory perception was found to be affected by shorter system delays then previously thought relevant. Human tracking data were obtained and used to develop a model that describes a subject's interaction with spatially dynamic sound sources.