data preprocessing and final loss value

[janeyeon]

Hi. I wanted to ask if you performed data normalization of an audio after trimming all the silences! And if you did, what method did you use? (maybe link to a paper or lecture or some package, please?)

What was final validation loss of G and P after training is almost done? My result is something like this.. and I'm not sure if it's an okay number.

[auspicious3000]

Please refer to the data preparation code for details

[steven850]

@janeyeon how where you able to train both G and P at the same time? willing to share your code?

[auspicious3000]

You don't need to train them at the same time.

[steven850]

Is there code available to train P, Im only seeing G in the solver.py

[auspicious3000]

You can simply replace G with P along with some other minor modifications.

[XintaoZhao0805]

Hi . Did you do some preprocessing such as trimming silence and cropping in the training data? Or Have you done some other changes in the code...? In my expeiment, the final validation loss is much higher than yours. I want to find the reason... Thank you in advance.

[auspicious3000]

All preprocessing steps are in the code, except trimming silence. But I don't think they will make any fundamental difference. Your loss value looks fine.

[FurkanGozukara]

@janeyeon @steven850 any ideas how to fix this problem?

thank you

https://github.com/auspicious3000/SpeechSplit/issues/28

[9527950]

Can you share this code of yours? Thank you very much!