Zhaoyu Zhang

Hi, thank you for your reply. I want to freeze the model from the TensorFlow checkpoint to .pb file. I am wondering which node is used for output the image...

@DiyuanLu How is your result of the final trained model? Does SILENCE_TRESHOLD = 0.1 work better than the original 0.3?

@weixsong Hi, what's the output node name that you used for freezing the model?

@Ostnie 1. You should use TensorFlow r1.1 rather than r1.3. TensorFlow moved rnn layers to tf.nn in r1.3. 2. change this folder '/home/pony/github/data/timit/phn/train/mfcc' to some folder you like in your...

@camer314 Is it possible to share the frozen model (.pb)?

Please note that the activation function used for "vad_gru" and "denoise_gru" in the C array weights is RELU rather than Tanh in the "rnn_train.py" script.

@zimenglan-sysu-512 Did you solve the problem?