Finite-difference computation of head-related transfer functions for human hearing

Abstract

Modeling the head-related transfer function (HRTF) is a key to many applications in spatial audio. To understand and predict the effects of head geometry and the surrounding environment on the HRTF, a three-dimensional finite-difference model has been developed for the head-related impulse response. A perfectly matched layer (PML) is used to absorb outgoing waves at the truncated boundary of an unbounded medium. An external source is utilized to decrease the computational domain size. This numerical model has been validated by analytical solutions for a spherical head model. The FD code is then used as a computational tool to predict the HRTF for various scenarios. In particular, a simplified spherical head model is compared to a realistic head model up to about 7 kHz. The HRTF is also computed for a realistic head model in the presence of a wall. It is demonstrated that this FD model can be a useful tool for spatial audio applications.