Fig. 1From: Sub-convolutional U-Net with transformer attention network for end-to-end single-channel speech enhancementProposed TANSCUNet model architectureBack to article page