Ambisonics Encoder for Wearable Array with Improved Binaural Reproduction

TL;DR

This paper proposes an improved Ambisonics encoder that combines Signal Matching and Binaural Signal Matching to enhance binaural reproduction accuracy for wearable microphone arrays, benefiting virtual and augmented reality sound experiences.

Contribution

It introduces a novel joint optimization framework integrating ASM and BSM, improving binaural reproduction quality from wearable Ambisonic arrays.

Findings

Enhanced binaural reproduction accuracy achieved

Joint ASM-BSM optimization outperforms previous methods

Simulation results demonstrate higher-quality virtual sound playback

Abstract

Ambisonics Signal Matching (ASM) is a recently proposed signal-independent approach to encoding Ambisonic signal from wearable microphone arrays, enabling efficient and standardized spatial sound reproduction. However, reproduction accuracy is currently limited due to the non-ideal layout of the microphones. This research introduces an enhanced ASM encoder that reformulates the loss function by integrating a Binaural Signal Matching (BSM) term into the optimization framework. The aim of this reformulation is to improve the accuracy of binaural reproduction when integrating the Ambisonic signal with Head-Related Transfer Functions (HRTFs), making the encoded Ambisonic signal better suited for binaural reproduction. This paper first presents the mathematical formulation developed to align the ASM and BSM objectives in a single loss function, followed by a simulation study with a simulated microphone array mounted on a rigid sphere representing a head-mounted wearable array. The analysis shows that improved binaural reproduction with the encoded Ambisonic signal can be achieved using this joint ASM-BSM optimization, thereby enabling higher-quality binaural playback for virtual and augmented reality applications based on Ambisonics.