A method for multi-source localization in real time using spherical harmonic decomposition

Abstract

By taking advantage of the spherical harmonic transform, spherical microphone arrays are able to decompose three-dimensional sound fields into their basic components, which enables detailed analysis and efficient spatial filtering algorithms. In recent years, methods for determining relative source positions around an array have been studied. Weighting functions can be generated to create a steerable beam which attenuates a sound field except in a specific direction. This research describes a real-time framework for multiple-source detection, which can be implemented using these methods. By finding the energy-sum distribution across a matrix of weighted harmonic signals, directional information may be simultaneously and efficiently obtained for multiple sources over time.