Fig. 2From: Speech dereverberation for enhancement and recognition using dynamic features constrained deep neural networks and feature adaptationThe architecture of the DNN for speech coefficient mappingBack to article page