Analysis of Frequency-Dependent Behavior of Room Reflections Using Spherical Microphone Measurements & Von Mises-Fisher Clustering

Abstract

This paper presents a room acoustic analysis tool capable of power response generation and directional characterization of room reflections across different frequencies using spherical microphone array measurements. The method exploits the spatial correlation between the frequency-dependent spherical harmonic coefficients of the reverberant soundfield and extracts its statistical features using von Mises-Fisher (vMF) clustering. We use this tool to examine the acoustic response of a small and a large room to achieve a profound understanding of the frequency-related variations in the directional characteristics of room reflections. In comparison to the eigen-beam multiple signal classification (EB-MUSIC) method, the proposed technique incorporates a more realistic room response over a broader frequency range. The experimental observations prove the potential of the proposed tool in determining the frequency-dependent room acoustic parameters and can lead to the design of smarter room acoustic treatment solutions.