Whisper-VITS-SVC
PlayVoice/whisper-vits-svcReviewed by AI
Checked by RepoPicks Radar on
What is Whisper-VITS-SVC?
An engine for singing voice conversion: you record a song, and it re-sings it with the voice of a singer whose model you trained from audio files. It also mixes voices to create a new one. It is aimed at people learning deep learning: the README says basic Python and PyTorch knowledge are required and that the authors will not make a one-click package. You need an NVIDIA graphics card (6 GB of video memory for training), Python, FFmpeg, several model files downloaded by hand, and a long list of commands. It does not work in real time.
Whisper-VITS-SVC is free and open source (MIT) and can replace Kits AI and Respeecher.
How easy is it to install Whisper-VITS-SVC?
D · terminal, several tools or a specific graphics card · Windows
- Windows D
- Linux D
There is no installer on any platform: on Windows and Linux alike you install Python and PyTorch, download five separate model files by hand (speaker encoder, whisper large-v2, HuBERT, crepe, pretrained model) and run many commands to prepare data, train and convert. Training needs an NVIDIA card with at least 6 GB of video memory, and the README states no one-click package will be made.
Coming soon: install in 1 click with the RepoPicks app.
Whisper-VITS-SVC on GitHub
- Stars
- 2,864on Nov 10, 2026
- Downloads
- 15,01033 this week
- Growth
- —no gain in the last reading
Stars over time
Facts
- Runs on
- Windows, Linux
- You need to install
- Python, NVIDIA graphics card, FFmpeg
- To build from source
- Python, NVIDIA graphics card, FFmpeg
- Disk space
- 150 MB–450 MB · estimate
- Accounts or keys
- None
- Cost
- Free · No paid service is needed; the models and code are free.
- License
- MIT (Permissive)
- Can replace
- Kits AI, Respeecher
- Last change on GitHub
- Apr 23, 2024
Whisper-VITS-SVC risk signals (7)
- Attentionstale releases
Last release on 2023-10-05 (1131 days ago)
automatic check
- AttentionDownload links to file-sharing sites
README links to https://drive.google.com/drive/folders/15oeBYf6Qn1edONkVLXe82MzdIi3O_9m3
automatic check
- AttentionDependencies with known vulnerabilities
Pinned dependencies with known vulnerabilities (OSV): gradio: GHSA-26jh-r8g2-6fpr, GHSA-279j-x4gx-hfrh, GHSA-34rf-p3r3-58x2
automatic check
- Attentionproject no longer updated
README.md says "No More Upgrade"; the last commit is from 2024-04-23 (930 days before the check).
AI review
- Attentionvoice cloning of real people
The repository description is "Core Engine of Singing Voice Conversion & Singing Voice Clone" and README lists "SNAC | Samsung | One Shot Clone of VITS"; copying a real singer's voice is restricted in some countries and may need the person's permission.
AI review
- Infothird party package mirror
README.md install step: "pip install -i https://pypi.tuna.tsinghua.edu.cn/simple -r requirements.txt" installs the Python packages from a mirror in China instead of the official index.
AI review
- Infomodels from file sharing and external sites
README.md asks you to fetch best_model.pth.tar from a Google Drive folder, large-v2.pt from openaipublic.azureedge.net and full.pth from another GitHub repository; nobody checks those files for you.
AI review
Evidence is shown in its original language.
What Radar could not check about Whisper-VITS-SVC
- Whether the five model downloads listed in the README are still available, since the project stopped being updated in 2024.
- Real disk use: the checks estimate 150-450 MB for the code, but the whisper large-v2 file alone is a few gigabytes and the prepared training data grows with your audio.
- Whether training really fits in 6 GB of video memory, and how many hours it takes on a normal home graphics card.
- Whether it works on Windows: the included environment.yml lists Linux packages only.
- The quality of the converted singing voice, which depends on your own recordings.