Fig. 3From: Text-independent speaker recognition based on adaptive course learning loss and deep residual networkThe calculation diagram of the CASP layerBack to article page