Fig. 4From: Discriminative frequency filter banks learning with neural networksNN-based structures with proposed methods. a is the NN structure for audio source separation tasks. b is the NN structure for audio scene classification tasksBack to article page