Perceptual consequences of interpolating head-related transfer functions during spatial synthesis

In implementing a spatial auditory display, many engineering compromises must be made to achieve a practical system. One such compromise involves devising methods for interpolating between the head-related transfer functions (HRTFs) used to synthesize spatial stimuli in order to achieve smooth motion trajectories and locations at finer resolutions than the empirical data. The perceptual consequences of interpolation can only be assessed by psychophysical studies. This paper compares three subjects' localization judgments for stimuli synthesized from non-interpolated HRTFs. Simple linear interpolations of the empirical HRTFs, stimuli synthesized from non-interpolated minimum-phase approximations of the HRTFs, and linear interpolations of the minimum-phase HRTFs. The empirical HRTFs used were derived from a different subject (SDO) from a previous study by Wightman and Kistler (1989) and whose data are provided with the Convolvotron synthetic 3D audio system. In general, the three subjects showed the same high rates of front-back and up-down confusions that were observed in a recent experiment using non-individualized (non-interpolated) transforms from SDO. However, there were no obvious differences in localization accuracy between the different types of synthesis conditions.<>