Acoustic reflection parameterization based on the Spatial Decomposition Method

The Spatial Decomposition Method (SDM) has gained interest within the acoustics community as a way to analyze and visualize the spatial sound field observed at a measurement position. Although the full visualization provides rich details of the sound field, it can be difficult to identify specific acoustic features or reflections within the analysis time frames. In this paper, we propose a parameterization of the SDM using a mathematical model of time-localized von Mises–Fisher distributions inspired by the image source model. Reflection directions of arrival, spatial concentration and diffuseness are estimated using an Expectation Maximization algorithm. Results are presented in our application domain of sound reinforcement. We compare parameters for sound reproduction with different acoustic absorption conditions and analyze how an immersive system in a theater excites the natural room acoustics.