Using a finetuned XTTS Model #27
-
|
Hi! I have trained a few XTTS models with the very good WebUI from @daswer123 (https://github.com/daswer123/xtts-finetune-webui) by myself. Is it possible to integrate the finetuned model into this WebUI? Because I think this WebUI is very good and the further range of functions is very interesting! I've seen that it's still on the to-do list to implement this feature. But is there perhaps a workaround that would have a solution for this by then? Thank you very much for the very good work so far and the help! Jonas |
Beta Was this translation helpful? Give feedback.
Replies: 14 comments 7 replies
-
|
Hi Jonas, I released a new version today and added support for fine-tuned models. Put your model files into the models folder (Pandrator/xtts-api-server/xtts_models/YOU_MODEL_NAME. There should be three files (model, vocab and config). Remember to also use a voice sample of the same voice you trained the model on, otherwise it won't work well. You can upload it from the GUI. Please let me know if it works well. All the best, |
Beta Was this translation helpful? Give feedback.
-
|
Hi! |
Beta Was this translation helpful? Give feedback.
-
|
Thanks, but a lot remains to be done :) How much audio did you use to train the models? Was it in English or another language? Are you happy with the results? |
Beta Was this translation helpful? Give feedback.
-
|
It's a German voice that I trained with XTTS. |
Beta Was this translation helpful? Give feedback.
-
|
Sorry for my silly question, but is there a better way to update the allready installed Pandrator then deleting the folder and running the new installer & launcher? :) 2024-09-09 19:33:00,959 - ERROR - An error occurred during RVC_CLI installation or prerequisites download: Prerequisites process exited with non-zero return code: 1 2024-09-09 19:33:00,960 - ERROR - Installation failed: Prerequisites process exited with non-zero return code: 1 2024-09-09 19:33:00,961 - INFO - Installation failed. Check the log for details. Thanks and best regards! |
Beta Was this translation helpful? Give feedback.
-
|
You mean to install a component (like RVC or another TTS engine)? Yes, it should be possible. I will check the RVC installation in case I broke something... |
Beta Was this translation helpful? Give feedback.
-
|
I had already installed Pandrator (with XTTS and RVC), but I wanted to install the latest version to integrate my finetuned XTTS models. |
Beta Was this translation helpful? Give feedback.
-
|
You can right-click in the Pandrator/Pandrator folder, open the terminal there, and run |
Beta Was this translation helpful? Give feedback.
-
|
Ah yes, now that I read it, it seems clear to me! I didn't think of that. |
Beta Was this translation helpful? Give feedback.
-
|
I don't know how I did it, but it seems like I shot myself git! |
Beta Was this translation helpful? Give feedback.
-
|
No, it was not related to git, it was missing a dependency for some reason. I've fixed it and added updating to the gui: https://github.com/lukaszliniewicz/Pandrator/releases/tag/v.0.0.9.5. Hope you won't have any major issues going forward :) |
Beta Was this translation helpful? Give feedback.
-
|
I experimented a bit again tonight. Unfortunately, I haven't had any success yet. But as soon as I put my finetuned model in the appropriate folder, XTTS doesn't start anymore. I've tried different things, like renaming the folder and files.
|
Beta Was this translation helpful? Give feedback.
-
|
Just to confirm: you have the three model files inside Pandrator/xtts-api-server/xtts_models/Folder_with_your_model_name? Could you perform a clean installation using the new installer, perhaps in a different folder if you don't want to delete the existing one? |
Beta Was this translation helpful? Give feedback.
-
|
Yes, I had copied the model with the folder for it from the folder structure of the XTTS Finetune WebUI von @daswer123 (https://github.com/daswer123/xtts-finetune-webui). There are five files in there. I then copied the complete folder into C:/KI/Pandrator/xtts-api-server/XTTS_models/VoiceName. I then renamed the VoiceName folder again and deleted the unnecessary files. An idea that came to me last night: maybe you had also changed something on the xtts-api-server!? I had done git pull in C:/KI/Pandrator/Pandrator. Maybe I should have done git pull in C:/KI/Pandrator? Probably yes... Otherwise, I would try again with a new installation and rename the Pandrator folder to Pandrator backup beforehand. That's what I did two days ago. I think it didn't work because I had shot myself to pieces. |
Beta Was this translation helpful? Give feedback.
-
|
So, I've definitely come a lot further.
Best regards! |
Beta Was this translation helpful? Give feedback.
-
|
Hmm, I can't replicate this, RVC works well for me. Does it happen for every model/index pair? Did you have any error messages during installation? Have you installed it using the latest version of the installer? |
Beta Was this translation helpful? Give feedback.
-
|
Okay, then it's probably my fault. I only installed Pandrator and then generated something as a test. Then I rebooted and installed XTTS, rebooted, and then installed RVC. When I generate with XTTS everything runs fine, but when I activate RVC it takes an extremely long time (over three minutes for four or five lines of text) and I have no result, even though a success is reported. I tested it with my XTTS model and with V2.0.2 in German and with a German wav file as a reference. |
Beta Was this translation helpful? Give feedback.
-
|
When I just started the Installer & Launcher, I noticed that Install RVC was selectable again. I then started the installation all over again. And I got an error message, the first error has to do with VS C++ Build Tools. This is the log file: 2024-09-12 19:52:24,254 - INFO - Installation process started. 2024-09-12 19:52:24,447 - DEBUG - STDERR: 2024-09-12 19:52:24,540 - DEBUG - STDERR: 2024-09-12 19:52:24,628 - DEBUG - STDERR: 2024-09-12 19:52:26,801 - ERROR - Installation failed: Command '['git', 'clone', 'https://github.com/blaisewf/rvc-cli.git', 'C:\KI\Pandrator\rvc-cli']' returned non-zero exit status 128. 2024-09-12 19:52:26,803 - INFO - Installation failed. Check the log for details. |
Beta Was this translation helpful? Give feedback.
-
|
So damn, I just remembered that I didn't run the I&L as an administrator! Log: 2024-09-12 20:01:16,590 - INFO - Installation process started. 2024-09-12 20:01:16,736 - DEBUG - STDERR: 2024-09-12 20:01:16,943 - DEBUG - STDERR: 2024-09-12 20:01:17,153 - DEBUG - STDERR: 2024-09-12 20:01:26,134 - ERROR - Installation failed: Command '['git', 'clone', 'https://github.com/blaisewf/rvc-cli.git', 'C:\KI\Pandrator\rvc-cli']' returned non-zero exit status 128. 2024-09-12 20:01:26,137 - INFO - Installation failed. Check the log for details. |
Beta Was this translation helpful? Give feedback.
-
|
Sorry, it's a bug in the installer, it doesn't remember or check that RVC is installed. You can remove the rvc-cli folder and try again. The Build Tools error is a bit strange, but you should have them installed already or XTTS wouldn't be installed. |
Beta Was this translation helpful? Give feedback.
-
|
Yes, i´ve had installed the Build Tools a while ago. I will delete the rvc-cli folder and try the installation again by time. |
Beta Was this translation helpful? Give feedback.
Hi Jonas,
I released a new version today and added support for fine-tuned models. Put your model files into the models folder (Pandrator/xtts-api-server/xtts_models/YOU_MODEL_NAME. There should be three files (model, vocab and config). Remember to also use a voice sample of the same voice you trained the model on, otherwise it won't work well. You can upload it from the GUI. Please let me know if it works well.
All the best,
Luke