Computationally Efficient Parametric Filter Approximations for Sound-Source Directivity and Head-Related Impulse Responses

Abstract

Simulation of virtual (room) acoustic environments (VAEs) has application ranging from architectural acoustics to entertainment and has become increasingly important in the context of ecological validity of hearing research and evaluation of hearing aid signal processing. Next to sound propagation and interaction with objects in the environment, the spatial directivity of the sound source and that of the receiver are important, including the human head and torso for headphone-based auralization. These directivities can be measured and represented by, e.g., head-related impulse responses (HRIRs). Particularly in interactive, real-time VAEs, considerable computational cost is required for interpolation and fast convolution with such directional filters. Approaches for simplifications exist including the approximation of the human head by a sphere. Here we suggest a unified structural approach to describe spatial directivity for sources and receivers including the human head with highly efficient low-order recursive filters. The filters are easily adjustable to match user requirements and are compared to the spatial directivity of the point source on a sphere and measured HRIRs. Perceptual evaluation showed good agreement with measured HRIRs. The suggested filter approximations are freely available as part of the room acoustics simulator (RAZR; www.razrengine.com).