You can deploy this model and start earning money today!

SmlCoke/
rvc-yui

mit

Views

No views yet

RVC-Yui Model Repository

English | 中文

This repository hosts Retrieval-based Voice Conversion (RVC) model weights (.pth) and index files (.index) for the character Hirasawa Yui (平泽唯) from K-On! (轻音少女). These models are open-sourced for community use.

📂 Repository Structure

.
├── index/                                          
│   ├── added_IVF1004_Flat_nprobe_1_yui-mix-hq_v2.index
│   ├── added_IVF1386_Flat_nprobe_1_yui-mix-pro-hq_v2.index
│   └── added_IVF1648_Flat_nprobe_1_yui-mix-pro_v2.index
├── weights/                                        
│   ├── yui-mix-hq.pth
│   ├── yui-mix-pro-hq-40k.pth
│   ├── yui-mix-pro-hq.pth
│   └── yui-mix-pro.pth
└── README.md

🏷️ Naming Convention

The filenames contain specific tags indicating the training configuration and dataset version used.

1. Dataset Version (`-pro`)

With -pro: Trained on v2 dataset. This version includes a larger dataset, combining ~10 classic songs from K-On! (v1 data) with additional voice data from the first season of the anime.
Without -pro: Trained on v1 dataset. This version only includes data from ~10 classic K-On! songs.

2. Data Quality (`-hq`)

With -hq: High Quality. Low-quality data (specifically .wav files labeled with "mid", containing weak short-term noise) has been removed from the dataset.
Without -hq: Standard dataset which may contain some "mid" quality segments.

3. Sampling Rate (`-40k`)

With -40k: Trained with a sampling rate of 40kHz.
Without -40k: Trained with the default sampling rate of 48kHz.

🔗 Links

RVC Project (Retrieval-based Voice Conversion WebUI): https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI
Dataset: https://huggingface.co/datasets/SmlCoke/rvc-yui-st

SmlCoke/rvc-yui