Create model repository

#1
by osanseviero HF staff - opened

Hey there! I see the model weights are hosted directly in the Space repository. Some other users have expressed interest in using this model, but as it's directly in this repo it's harder to be used.

Would it be possible to move the model weights to a model repository? You can then programmatically access the model files in this space using the huggingface_hub library (here are some docs on how to achieve it).

One of the advantages of keeping them separate is that you can add a model card that documents how the model can be used, its intended use cases and which are its biases.

i second that it would be great, @BilalSardar ! let us know if we can help in any way

Original model files are here:
https://github.com/Edresson/YourTTS
@osanseviero @julien-c
This space is an adaptation of YourTTS voice conversion. Normally i don't have issues with adaptations but i think this guy should have cited that work.

If you want to use this space as API. You can load it by using gr.Interface.load(name="spaces/BilalSardar/Voice-Cloning") function and if you want models details you can visit Zero-Shot TTS . @ugtgl thanks for sharing. @julien-c Thanks for offering help buddy.

This comment has been hidden

Sign up or log in to comment