My main project
Ilaria RVC
Ilaria RVC is my Windows app for voice conversion and training. I built the current version on Ilaria RVC Mainline, which grew out of the RVC Project’s WebUI and contributions from the community. The interface runs locally in a browser. It currently supports NVIDIA GPUs only.
- Voice conversion
Apply a model to one recording or a batch. You can also start with text, generate speech and run it through a voice model in the same app. RMVPE and FCPE extract pitch so the converted voice can follow the original melody or intonation.
- Training
Check the dataset before starting, follow progress live and listen to previews as the voice takes shape. The training queue lets you line up several jobs. A built-in Audio Analyzer also helps you inspect your recordings.
- Model management
Download, import, inspect, merge and export voices from the Other tab. The 1.9 series added streaming local imports for ZIP, PTH and INDEX files with progress, cancellation and duplicate checks, plus an offline changelog. The Windows package and its updates are on Hugging Face.
To try it, extract the Windows package and run ILARIA_RVC.bat. Import a voice model, choose an audio clip and listen to the converted result. For a model you want to train, start with a dataset of recordings. You can check for app updates from the Other tab.