steven

This may caused by webrtc bugs, I have post this on webrtc discuss group here:https://groups.google.com/forum/#!topic/discuss-webrtc/GqPl8rKpt7Q

This may caused by webrtc bugs, I have post this on webrtc discuss group here:https://groups.google.com/forum/#!topic/discuss-webrtc/GqPl8rKpt7Q

I think 'DTLN-aec' include 'Noise suppression' feature.You can read the paper of 'DTLN-aec'.

> @yudashuixiao1 Mybe the reason is the code : early_stopping = EarlyStopping(monitor='val_loss', min_delta=0, patience=10, verbose=0, mode='auto', baseline=None) when you val_loss did not improve ,large than patience=10 epoch, training will be...

> @steven8274 hello! may I ask how to understand the "All signals used as input to the model are subject to a random gainchosen from a uniform distribution ranging from...

> @steven8274 thank you for you reply , I still don't quite understand what signal to add，example： > > import numpy as np min_gain_dB = -25 max_gain_dB = 0 random_gain_dB...

Maybe you used an incompatible CUDA with your GPU.Check which versions of CUDA are compatible with your GPU, install the version which is compatible with the closet version of 2.2.0...

I found that this problem only occured when input image's resolution is under 2K, e.g. 1920x1080, 1440x900, etc.MUST the input image's resolution be at least 2K?

I alse found that IRN_x2 model does not cause such problem.Besides, IRN+_4x model cause such problem too.

> Probably because the model is only trained on high-resolution images (DIV2K) and does not generalize very well for some unseen distributions of input images. For real applications, you may...