On using time-frequency binary masking for dereverberation

The paper presents some aspects regarding the recent introduced time-frequency (T-F) binary mask for speech dereverberation. First of all, a theoretical analysis of the binary mask is presented. It is compared with another T-F binary mask which was used in speech separation, to find some common aspects but also the differences. By using the results of this comparative analysis, we demonstrate the existence of the T-F binary mask for speech dereverberation. To prove our results, relevant experiments are presented.