Abstract:
We propose a method for estimating the direct-to-reverberation energy ratio (DRR). With this method, diffuse sound field is assumed for reverberation, i.e., reverberation is modelled to propagate into a microphone isotropically with a homogeneous power spectrum density (PSD).With this model, the proposed method estimates the PSD of both direct sound and reverberation by exploiting the directivity shape of a beamformer. Simulation under an ideally diffuse field proved that the proposed method can accurately estimate the DRR from - 20 to 20 dB.