Testing listeners' ability to localize virtual sources using sparse loudspeaker arrays and comparison with spherical-head model predictions

Virtual acoustic environments are created by mimicking pressures of real acoustic sources at the ears of a listener. One method to control these pressures is through the use of loudspeaker arrays. While the best results come from arrays with a large number of loudspeakers, they are not practical for everyday use. On the other hand, sparse loudspeaker arrays are physically limited by the listener's localization ability and the sensitivity to head translation and rotation. Previously a model was created to find the limits of each of these features for a spherical head in a free-field environment. Subsequently the loudspeaker locations were optimized based on these features. Listening tests were carried out to find the limits of these features for a variety of loudspeaker array configurations. These tests were carried out in an anechoic chamber with the loudspeaker locations hidden by acoustically transparent curtains. The localization ability for each configuration was recorded. This was repeated for deviations from the optimal listener location in both head rotation and translation. The results were compared to the free-field model previously established.