Optimization of loudspeaker positions for virtual auditory displays in a free-field environment

Abstract

Virtual acoustic environments are created by mimicking the pressures of real acoustic sources at a listener’s ears. In this way, a listener can perceive virtual acoustic sources as coming from any desired direction. There are multiple methods to accomplish this using headphones or loudspeaker arrays, but not all are practical for everyday use. Some use cases limit the use of headphones or the number and position of loudspeakers, such as automobiles and home theater systems. These limitations also put bounds on the performance of each possible system, such as allowable head movement and the accuracy of localization. An exhaustive-search optimizer was used to find the performance of various sparse loudspeaker arrays, given a limited number of possible loudspeaker positions in a free-field environment for a single stationary listener. The optimizer sought to maximize allowable head rotation and translation and to minimize localization errors. Arrays consisting of two or four loudspeakers were considered using a spherical head model to find the necessary head-related transfer functions and cross-talk cancellation was used to create the virtual acoustic environment. From the results, pareto fronts were created that can be used to find the best loudspeaker array configuration for given design constraints.