A Parallel Approach to HRTF Approximation and Interpolation Based on a Parametric Filter Model

Abstract

Abstract-Spatial audio rendering techniques using headrelated transfer functions (HRTFs) are currently used in many different contexts such as immersive teleconferencing systems, gaming, or 3D audio reproduction. Since all these applications usually involve real-time constraints, efficient processing structures for HRTF modeling and interpolation are necessary for providing real-time binaural audio solutions. This letter presents a parametric parallel model that allows to perform HRTF filtering and interpolation efficiently from an input HRTF dataset. The resulting model, which is an adaptation from a recently proposed modeling technique, not only reduces the size of HRTF datasets significantly, but it allows for simplified interpolation and real-time computation over parallel processors. In order to discuss the suitability of this new model, an implementation over a graphic processing unit (GPU) is presented.

An Efficient Implementation of Parallel Parametric HRTF Models for Binaural Sound Synthesis in Mobile Multimedia

“…The model, implemented as a SOS chain, was first proposed in [19], demonstrating the benefits of the parametric approach in terms of data storage needs and interpolation simplicity. Later, a parallel adaptation was developed in [20] to overcome the serial-to-parallel conversion while maintaining the desirable properties of the original model. For the sake of a self-contained presentation, the next subsections briefly describe both models, with emphasis on the parallel one as it is the final model optimized, implemented and evaluated in this paper.…”

“…To illustrate this problem, an example extracted from [20] is shown in Fig. 5(a), where the red line indicates the result of the parallel model by PFE obtained from the sequential model with N = 6.…”

“…As previously discussed, while PFE provides a parallel equivalent of the sequential model that can be efficiently implemented on a GPU, synthesizing new responses from modeled directions by direct interpolation of the parameters is not possible. To solve this issue, the authors proposed in [20] a parallel version for HRTF modeling following the same parametric philosophy of the serial implementation. Parallel filter bank implementations in audio applications are common [28]- [31], being one of their benefits their better signal-to-noise ratio properties [32].…”

“…This is because of the relevance of the phase relation between the parallel SOS, which has no impact on the serial model. Nonetheless, both models have a similar computational cost [20].…”

“…Some methods employ an analytical model of the head and torso, modeling the propagation delay and the diffraction effects of the wave. Other approaches are based on the design of a digital filter that models the behavior of the original HRTF, such as the one implemented in this paper that was previously developed by the authors [19], [20], based on infinite impulse response (IIR) filters. As will be seen, the HRTF model implemented in this paper has several advantages that makes it suitable to be executed in multimedia mobile devices.…”