How to use Custom Voice Models

[feijoes]

Is there any way to use my custom voice model , for example any from https://voice-models.com/ ?

[harry0703]

cool
are there any APIs?

[feijoes]

@harry0703
I'm not entirely sure, but I believe it would be beneficial to implement the feature where users can specify a link for the location of the dowloaded zip file containing the custom voice model. This would allow for easier integration and usage of personalized models.

[feijoes]

Maybe you can use https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI or others similars projects to convert the voice from edge-tts with the user custom model , If you think it's a good idea, I can fork the repo and work on it myself.

[feijoes]

hey @harry0703 I'm working on implementing the custom model feature. Do you have a Discord/email or any other means of communication for updates or questions? Or should I just address them here in this issue?

[harry0703]

Thank you very much.
We can communicate and exchange ideas directly here.
I will check and reply in a timely manner.

[9xcoder]

Maybe you can use https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI or others similars projects to convert the voice from edge-tts with the user custom model , If you think it's a good idea, I can fork the repo and work on it myself.

Can you introduce some projects or simple ways to use pre-trained models from the above website?

[feijoes]

@9xcoder What you mean ? Here is a tutorial of how to use rvc correctly https://www.youtube.com/watch?v=hSxTLCR_95Y

[feijoes]

Hey @harry0703, here's an update on the issue. I've added the submodule https://github.com/skshadan/TTS-RVC-API to this project and removed all unnecessary parts. For now, it's working as an external API. My plan is to integrate it, not just as an external API on another port, but to integrate it with others. Additionally, I'm planning to add a button in the GUI interface that allows users to input the link of the external model, download it, extract it into the correct folder, and then select it as a speaker. However, I've encountered some issues with this. With the inclusion of this project, the project's requirements have changed significantly because all the audio conversion processes with a custom model are now done by the computer without any API. For example, the Dockerfile build now takes 30 minutes on my PC, even after removing all unnecessary files. Perhaps it's a good idea for this to be an optional feature of the project.

[harry0703]

@feijoes This is a great idea, as some users have similar needs, and it can be flexibly configured as an optional feature.
Thank you for your contribution, really looking forward to it.

[9xcoder]

@feijoes
Sorry for my poor English skills.
Actually, I want this project (TTS-RCV-API) to be able to integrate into my other project.
Thank you for the YouTube link, it provides me with a method to train the model.

[feijoes]

@9xcoder Sure, follow the readme of the project and this should work , maybe you would like to remove some of the unnecessary files, but this project has all the minimum requriments to run it, if you need any help , please let me know