Perceived stability and localizability in binaural reproduction with sparse HRTF measurement grids

With the increased popularity of virtual reality applications, the need for high fidelity spatial audio has emerged. Reproduction of high quality spatial audio requires high resolution individualized head-related transfer functions (HRTFs). However, these are typically unavailable as they demand a large number of measurements and specialized equipment. Given sparse measurements, it is necessary to spatially interpolate the HRTF before employing it in binaural reproduction. Prior studies suggested the use of spherical-harmonics (SH) representation as the basis for the interpolation. However, interpolation of sparse measurements may lead to errors due to spatial aliasing and SH series truncation. In this study, the effects of these errors on the perceived acoustic scene stability and on the localizability of the virtual sound source are investigated numerically, as well as perceptually. Experimental results indicate a significant effect on both attributes due to truncation error, while the effect of the aliasing error is less dominant.