From RIR to BRIR: A Sparse Recovery Beamforming Approach for Virtual Binaural Sound Rendering

Abstract

The creation of a spatial sound scene through binaural rendering draws increasing research interest, given the rising application of virtual reality and augmented reality. High-fidelity augmented reality audio requires accurate room acoustic simulation. Typically, this is achieved with binaural rendering of spatial sounds using recording of a Higher-Order Microphone (HOM) and Head-Related Transfer Functions (HRTFs) based simulations. However, HOM often fails to deliver full immersion due to the limited order of recording. In this paper, we introduce a Binaural Room Impulse Response (BRIR) calculation method based on HOM recordings of Room Impulse Responses (RIRs) to improve binaural sound rendering. The proposed method enables a more accurate analysis and rendering of the directive components in the limited-order spherical harmonic (SH) recording of RIRs with sparse recovery beamforming. Results with measured RIRs using a HOM indicate that the proposed method achieves a more accurate BRIR calculation than the conventional SH domain method.