Dereverberation using a model for the spatial coherence of decaying reverberant sound fields in rectangular rooms

摘要

In rooms with nonuniform absorption due to different material properties, the spatial properties of the decaying reverberant sound field were found to be time-variant, violating the common assumption of a diffuse sound field. Recently, we proposed a model for the coherence of such a sound field in a rectangular room, using knowledge of the geometry and surface reflection coefficients. In this paper, we describe a coherence-based dereverberation method based on this model. We conduct an evaluation with impulse responses which were generated for rooms with non-uniform absorption, confirming that the model can be applied for speech dereverberation.