A STUDY ON SOUND SOURCE APPARENT SHAPE AND WIDENESS

This work is intended as an initial investigation into the perception of wideness and shape of sound sources. A method that employs multiple uncorrelated point sources is used in order to form “sound shapes”. Several experiments were carried out in which, after some initial training, subjects were asked to indentify the shapes that were being played. Results indicate that differences in vertical and horizontal source wideness are easily perceived and scenes that use broad sound sources to represent normally large sound objects are selected 70% of the time over point source versions. However, shape identification was found to be more ambiguous except for certain types of signals where results were above statistical probability. The work indicates that shape and wideness of sound sources could be effectively used as extra cues in virtual auditory displays and generally improve the realism of virtual 3D sound scenes. This work was performed as a Core Experiment within the MPEG Audio Subgroup with the intention of possible integration of source wideness into MPEG-4 AudioBIFS.