Fig. 6From: Channel and temporal-frequency attention UNet for monaural speech enhancementThe learned attention matrix of each layerBack to article page