Fig. 3From: Progressive loss functions for speech enhancement with deep neural networksFront-end. Preprocessing block at beginning of P-CNN and P-ResNet architecturesBack to article page