Efficient Binaural Rendering of Moving Sound Sources Using HRTF Interpolation

Abstract

This paper presents efficient techniques for the simulation of continuously moving sound sources, to be presented over headphones, based on the interpolation of Head-Related Transfer Functions (HRTFs). Two novel approaches are presented, one based on interpolating impulse responses in time-domain, and another based on interpolating poles and zeros of low-order Infinite Impulse Response (IIR) approximations of measured HRTF filters. These techniques are apt both for real-time implementation and for multi-user applications where several independent binaural renderings are required. We discuss some of the difficulties of assessing the quality of the results from both objective and subjective points-of-view, and provide objective measurements of the errors caused by the interpolation schemes, as well as of the errors introduced by the use of low-order IIR approximations. A subjective evaluation of these techniques using non-individualized HRTFs suggests that they are able to preserve spatial impression of directions in binaural rendering of static and moving sound sources.

Acknowledgements

The authors would like to express their gratitude for the support obtained from CAPES, CNPq and FAPESP for this research.