How to make an AI cover using Hugging Face

Guide made by Julia

Credits to the user r3gm for the Hugging Face space.

( You can DM me on Discord if you have any question or encounter any issue: ailen2091 )

Notes: Works on mobile too. It's ideal if you have a slow device and you can't run RVC locally, since this software doesn't use the resources of the device itself in order to work. Just make sure you have a stable Internet connection.

-First make an account in Hugging Face ( https://huggingface.co/join ).

-Once you are logged in go to the space ( https://huggingface.co/spaces/r3gm/Aesthetic_RVC_Inference_HF ), click on the three dots and "Duplicate this Space". YOU MUST DUPLICATE THE SPACE OR ELSE IT WILL GIVE YOU AN ERROR. Make sure you set it to Private and then hit Duplicate.

It will take a few minutes for it to load.
If you want to use the space again, you don't have to duplicate it again. Your duplicated spaces will be in your profile. On the upper right corner you click your icon and then your username. Your list of spaces will be there.

Note: if you don't use the space for a few days, by default it will go on "sleep" mode. That's normal. Just open it and click Restart this space and wait for it to load.

-Once it finishes loading, go to the Resources tab and paste the link of the model and click Download. It has to be a link from either Hugging Face, Mega or Google Drive, or else it will give you an error. You can either do that or manually upload the .index and .pth file, right under Download.

-Then go back to the main tab and click Refresh and select your model in "Inferencing Voice" and "Auto-detect index path".

-Below it, select your audio file. If you want to convert multiple audios at once, you can do it in the Batch tab.

-Set up the pitch and the settings in Advanced Settings however you want, and you are good to go.

-Click Convert and wait. Keep in mind, if you leave the page for a long time it might lose the connection and it will give you an error. I recommend to come back to the page a few times while it renders.

-To download it the output audio, you click on the three dots on the right of the audio and then Download.

Note: The conversion won't be instantaneous. A 1 minute-long audio will take around 5 minutes. If you want to make a faster space, go to the original space, click on the three dots, Duplicate this Space, and in "Space hardware" you select another hardware. Keep in mind that it's paid and you'll be charged per hour.