pyVideoTrans
jianchang512/pyvideotransReviewed by AI
Checked by RepoPicks Radar on
What is pyVideoTrans?
pyVideoTrans takes a video or audio file, writes down what is said, translates it, reads the translation with a synthetic voice and puts the new audio and subtitles back on the video. It can tell speakers apart, give each one a different voice, clone a voice from a sample and let you correct the text at each step. It can work offline with local models or connect to online services such as OpenAI, Azure, DeepL, Gemini and ElevenLabs. On Windows 10 or 11 there is a ready package to unzip and run; on macOS and Linux the README describes installing from source.
pyVideoTrans is free and open source (GPL-3.0) and can replace Rask AI, Descript, and HeyGen.
How easy is it to install pyVideoTrans?
B · installer plus one simple step · Windows
- Windows B
- macOS D
- Linux D
Windows 10 and 11 get the best grade: the README's quick start is to download the pre-packaged version from the releases page, unzip it to a folder without spaces and double-click sp.exe, then pick a recognition model or enter a key for an online service inside the app. For macOS and Linux the README only describes the source route — Python 3.10, FFmpeg installed by hand with brew or apt, the uv tool, `uv sync` and `uv run sp.py` — or a Docker image you build yourself.
pyVideoTrans on GitHub
- Stars
- 19,697on Nov 10, 2026
- Downloads
- 220,330all releases
- Growth
- +62stars in 6 days
Stars over time
Facts
- Runs on
- Windows, macOS, Linux
- You need to install
- Nothing extra
- To build from source
- Python, Docker, NVIDIA graphics card, FFmpeg
- Disk space
- 70 MB–210 MB · from the installer
- Accounts or keys
- OpenAI, Azure Speech, ElevenLabs, DeepL, Google Gemini, DeepSeek
- Cost
- Free · The program is free (GPL-3.0) and can run with free pieces (Edge-TTS, local Whisper); the optional online services it connects to, such as OpenAI, Azure or ElevenLabs, are paid
- License
- GPL-3.0 (Copyleft)
- Can replace
- Rask AI, Descript, HeyGen, ElevenLabs
- Last change on GitHub
- Nov 10, 2026
pyVideoTrans risk signals (6)
- AttentionInstall command runs a script from the internet
Install command pipes a remote script into a shell: curl -LsSf https://astral.sh/uv/install.sh | sh
automatic check
- Attentionvoice cloning impersonation risk
README feature list: "Voice Cloning: Integrates models like F5-TTS, CosyVoice, GPT-SoVITS for zero-shot voice cloning", which can copy a real person's voice from a sample
AI review
- Attentiondependency conflicts overridden
pyproject.toml pins about 250 packages and then adds a block commented "# 忽略版本冲突错误" (ignore version conflict errors) with override-dependencies, plus `index-strategy = "unsafe-best-match"`, which lets the installer pick a package from any of the configured indexes
AI review
- Infodocker build downloads third party binary
The Dockerfile downloads an FFmpeg build from a personal site during the build: wget -q https://johnvansickle.com/ffmpeg/releases/ffmpeg-release-amd64-static.tar.xz
AI review
- Infoauthor disclaimer about copyrighted video
README Disclaimer: "Users are solely responsible for any legal consequences arising from the use of this software (including but not limited to calling third-party APIs or processing copyrighted video content)"
AI review
- Inforeadme promotes paid sponsors
The README opens with a "Sponsors" table advertising Recall.ai and OfoxAI, both paid API services, with utm tracking links
AI review
Evidence is shown in its original language.
What Radar could not check about pyVideoTrans
- Which files the Windows package contains and whether local models come with it or download on first use
- Whether the Mac files on the releases page are a ready app, since the README documents only the source route for macOS
- Real disk use and video memory after downloading recognition, translation and voice models
- The quality of the automatic dubbing and lip timing for a given language pair
- How much the optional online services cost for a long video