Fig. 3From: Components loss for neural networks in mask-based speech enhancementProposed CNN training setup for speech enhancement according to the white-box approach. The hereby applied components loss (CL) is given in (8) and (11)Back to article page