What the Spatial Decomposition Method can and cannot do

Abstract

The spatial decomposition method (SDM) is a parametric approach for the processing of spatial room impulse responses. Although extensively used, there are some issues related to used microphone array, reproduction loudspeaker setup, and signal processing that are not widely understood. For example, there have been different observations with regards to its performance with rendering extremely short transient signals, where some authors have described issues such as "roughness" or "graininess". Here, we wish to clarify what the limitations of the spatial decomposition method are and provide practical guidance to ensure the value of SDM as a tool for spatial room impulse response analysis and rendering in various contexts. Specifically, we discuss directional estimation performance, the ability to estimate two reflections at the time, estimation in the late tail of the response, and the roughness and whitening found in the quality of the final rendering. Finally, some post-processing techniques to compensate the possible audible artifacts are reviewed.