This module is a synchronized version of Batch Normalization when using multi-gpus for deep learning, aka 'Syn-BN', as the mean and standard-deviation are reduced across all devices during training.
path_dnn_noisy_train = os.path.join(PATH_ROOT, args.input_noisy_train) dnn_magnitude_noisy_train,_,sr = perform_stft(path_dnn_noisy_train, args) # dnn_magnitude_noisy ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results