Open RECKENEFIN opened 1 year ago
Because WebUI is udpated by me, and colab is updated by some other persons. Somebody to pr the updated colab version is needed.
Got it. Well, until there's an official update, I did some tinkering and got it working in a copy of the v2 Colab Notebook.
torchcrepe
to the Dependencies listRMVPE.pt
to the Weights Download list I ran a small model through it and everything seemed to work fine. A few DEBUG notifications in the terminal during training, but nothing that ground things to a halt. I'm sure there are some back end improvements that would streamline it a bit more, but it's functional enough for me right now.
Got it. Well, until there's an official update, I did some tinkering and got it working in a copy of the v2 Colab Notebook.
- Added
torchcrepe
to the Dependencies list- Updated Clone Repository to fetch at Commit 56917db (from the day of the latest version release)
- Added
RMVPE.pt
to the Weights Download list- Enabled all the v2 weights (since those all seem to be functional now) and disabled all the v1 & v2 32k weights (since there isn't a 32k option in the current webUI).
I ran a small model through it and everything seemed to work fine. A few DEBUG notifications in the terminal during training, but nothing that ground things to a halt. I'm sure there are some back end improvements that would streamline it a bit more, but it's functional enough for me right now.
Hi, can you give me link please? Thanks
Sure, no problem. I'm new to the whole Github thing when it comes to etiquette and whatnot, and I know just enough about coding to be dangerous.
Retrieval_based_Voice_Conversion_WebUI_v2_RMVPE.txt
(Obviously change the file extension back to .ipynb
. Couldn't attach that file format.)
Sure, no problem. I'm new to the whole Github thing when it comes to etiquette and whatnot, and I know just enough about coding to be dangerous.
Retrieval_based_Voice_Conversion_WebUI_v2_RMVPE.txt
(Obviously change the file extension back to
.ipynb
. Couldn't attach that file format.)
HI, I change colab code as same as "Retrieval_based_Voice_Conversion_WebUI_v2RMVPE.txt" you provide but I find the issue is that !cp /content/Retrieval-based-Voice-Conversion-WebUI/logs/{MODELNAME}/added.index /content/drive/MyDrive/ !cp /content/Retrieval-based-Voice-Conversion-WebUI/logs/{MODELNAME}/total_.npy /content/drive/MyDrive/ doesn't work. Also the main function don't produce ".index file" and ".npy file". How to adjust that?
Sure, no problem. I'm new to the whole Github thing when it comes to etiquette and whatnot, and I know just enough about coding to be dangerous.
Retrieval_based_Voice_Conversion_WebUI_v2_RMVPE.txt
(Obviously change the file extension back to
.ipynb
. Couldn't attach that file format.)
Thank you for your work. Is it working smoothly so far, for the RMVPE part?
So far so good. Haven't had any issues with it thus far.
I just updated RVC on my local setup (which is only powerful enough to handle inference) and noticed the RMVPE option. I wanted to try it out on the training side, but the Colab doesn't appear to have that functionality enabled. I've noticed that the Colab does seem to be behind what the local version has in terms of features (no 48k v2 support, no Crepe option, no RMVPE option, etc.). Is there a specific reason for this, or am I missing something?