Fundamentals of a parametric method for virtual navigation within an array of ambisonics microphones

Abstract

Fundamental aspects of a method for virtual navigation of a sound field within an array of ambisonics microphones, wherein the subset of microphones to be used for interpolation is determined parametrically, are presented. An existing, weighted-average-based navigational method serves as a benchmark due to its simplicity and its applicability to arbitrary sound fields but introduces comb-filtering and, for near-field sources, degrades localization. A critical review of existing methods is presented, through which a number of issues are identified. In the proposed method, those microphones that are nearer to the desired listening position than to any source are determined based on the known or inferred positions of sources. The signals from only those microphones are then interpolated using a regularized least-squares matrix of filters. Spectral distortions and source localization errors are characterized for the benchmark and proposed methods via numerical simulations of a two-microphone array, and an experimental validation of these simulations is presented. Results show that, for near-field sources, the proposed method significantly outperforms the benchmark in both spectral and localization accuracy due to the exclusion of the second microphone. For far-field sources, the proposed method achieves slightly decreased spectral distortions due to the flattened response of the interpolation filters.