Haoran Zhou

I didn't spend too much effort fine tune the model when I wrote this piece of code. I just found that this lr work for the data set and the...

Sorry but it exceeds GitHub's file size limit of 100.00 MB.

Thank you for your great advise. I am now not able to access my labs' server because it's Winter Vocation in China and I am not at my school. After...

Sorry about the delay, I just used the number of validation samples in my validation set, of course you can set the number to a new value once you set...

Glad to know you like it. They are directorys for speech and noise used during training and cross validation.

There is a free dataset online with 100 types of noise (sorry I forget its name), and it can be used to test the efficacy of CNN for enhancement. A...

This error is probably due to the versions of the dependencies, and I guess rounding the float number to an integer will get things to work.

@imchukwu Thanks for finding out the reason. The training time depends on the types of dataset you use. I used TIMIT and 100 types of noise generating a dataset more...

Original audio files will work On 11/16/2018 21:01, WHUers wrote: Excuse me，the 4 fold :speech dir,noise dir, val _speech_dir,val_noise_dir ，in the 4 folds contain the original audio files or the...

All the data reading threads are started at the beginning and will get automatically filled during training loop