Comparison of Simulated Head-Related Transfer Functions Accuracy for Different Model Complexities Using the Finite-Difference Time-Domain Method

Abstract

The use of finite-difference time-domain (FDTD) simulations is relevant for several applications in virtual acoustics. One of these is the numerical calculation of headrelated transfer functions (HRTFs). This study investigates the effect of varying the geometrical complexity (shape, level of details) of a human head/torso model on the calculation of its HRTFs using an FDTD solver. In particular, the interest is on the accuracy of the obtained simulation results with respect to the human head/torso model complexity. For that aim, a solution verification process is undertaken, and a single sphere, a two-sphere and a human head and torso models are considered. The results indicate that relatively small 95% confidence intervals on the solution verification results are achieved, indicating relatively good accuracy for the prediction of HRTFs up to relatively high frequencies for the single and two-sphere models considered. However, for the simplified human head and torso model, a similar accuracy is achieved only up to a lower frequency.