New Windows users: Get and install portable version from original repo: https://github.com/Tiger14n/RVC-GUI/releases/tag/Windows-pkg

To drastically speed up subsequent conversions of a single input file using the same f0 method, simply replace rvcgui.py and vc_infer_pipeline.py in installation directory with the updated ones from this fork.

f0 reusing (caching) works across different voice models and pitch shifts. Cached f0 calculations are stored alongside the input file with the filename suffixed with used method (e.g. "input_file_name_harvest.f0", "input_file_name_crepe128.f0" etc, where the number represents hop value for crepe)

RVC GUI

For audio file inference only

GUI

Direct setup for Windows users

Windows-pkg

Preparing the environment

• Install Python version +3.8 if you have not:

• Execute these commands

Windows with Nvidia cards

python -m pip install -U pip setuptools wheel
pip install -U torch torchaudio --index-url https://download.pytorch.org/whl/cu118
pip install -r requirements.txt

Other

python -m pip install -U pip setuptools wheel
pip install -U torch torchaudio 
pip install -r requirements.txt

Apple silicon Macs fix

pip install --pre torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/nightly/cpu

export PYTORCH_ENABLE_MPS_FALLBACK=1

• Downlaod hubert_base.pt and place it in the root folder

• Then use this command to start RVC GUI:

python rvcgui.py

Or run this file on windows

RVC-GUI.bat

Loading models

use the import button to import a model from a zip file,

• The .zip must contain the ".pth" weight file.
• The .zip is recommended to contain the feature retrieval files ".index"

Or place the model manually in root/models

models
├───Person1
│   ├───xxxx.pth
│   ├───xxxx.index
│   └───xxxx.npy
└───Person2
    ├───xxxx.pth
    ├───...
    └───...

How to get models?.

• Join the AI Hub Discord
• Community Models on HuggingFace by Wicked aka QuickWick

K7#4523