Common Image Source Search for Multiple Spatial Room Impulse Response Measurements

Abstract

Image source reversion algorithms estimate room geometry from measured spatial room impulse responses by locating image sources. However, most of the methods have been limited to a single loudspeaker position and to convex rooms. Earlier, we have proposed a method that combines image sources from multiple receiver locations to find more image sources accurately even in concave rooms. Here, we extend the method to cope with multiple sound sources, thus the image source search can utilize measurements from multiple source and receiver positions simultaneously. The search method is tested with two measurement datasets and is found to improve the result compared to the previous algorithm. For the future applications, we also propose methods for generating a concave room model from the found reflection planes, for estimating material filters for each surface, and for interpolating between measured source locations.