Abstract:

The experimental method of Hartmann and Wittenberg [J. Acoust. Soc. Am. 99, 3678--3688 (1996)] is an approach to virtual reality with many limitations, but it is the most reliable technique for scientific purposes. Experiments using this method find that a precise VR synthesis is entirely convincing. Of interest are the perceptual consequences of perturbations on the precise synthesis. Large, random perturbations lead to diffuse and internalized images. Better controlled perturbations can lead to split images with perceived timbre differences and even pitch differences. Incorrect images overwhelmingly tend to be localized in the back, e.g., reversing interaural time differences while maintaining interaural intensity differences reverses the azimuth but also reflects the image in a front--back plane (when not internalized). Smaller perturbations produce images that are unitary, but not compact, and they are normally located at distances that are nearer than the true source. Perturbations of intermediate size can lead to extended but well-defined source images, such as gobs of sound extending from the back of the neck and running down the spine. Spatial effects like these can be disconcerting and illustrate the perils of synthesis, getting it almost right, but not entirely right. [Work supported by the NIDCD, DC00181.]