Spatial perception of motion-tracked binaural sound

Motion-tracked binaural sound reproduction extends conventional headphone-based binaural techniques by providing the dynamic cues to sound localization produced by voluntary head motion [V. R. Algazi, R. O. Duda, and D. M. Thompson, J. Aud. Eng. Soc. 52, 1142–1156 (2004)]. It does this by using several microphones to sample the acoustic field around a dummy head, interpolating between the microphone signals in accordance with the dynamically measured orientation of the listener’s head. Although the provision of dynamic cues reduces the sensitivity of the method to characteristics of the individual listener, differences between the scattered field produced by the dummy head and the scattered field that would be produced by a particular listener distorts the spatial perception. A common observation is that sound sources appear to rise in elevation when the listener turns to face them. We investigate this effect by comparing the perceived rise in elevation under three different conditions: recordings in which recordings are made using (a) the listener’s own head, (b) a KEMAR mannequin, and (c) a cylindrical head with no torso. Quantitative results are presented showing the degree to which perceptual distortions are least for (a) and greatest for (c). [Work supported by NSF.]