Win10/11 Pre-packaged Version Download

The current latest version is v3.81, and the patch package is v3.81.

Support the Developer, Help Maintain the Project

Thank you for using pyvideotrans. This is an open-source, free software developed by an individual developer in their spare time.

The project is not commercially operated and has no paywalled features or built-in ads. If you appreciate this project and wish for its continued development, consider supporting us with a small donation. Every bit of support is motivation for the project to continue.

If you encounter issues, besides checking this documentation, you can also ask questions in the online Q&A community.

Users who have donated can get problem support via email: [email protected]

v3.81 Full Package Download (2.5G)

Update Notes for v3.81

1. Added VoxCPM-TTS dubbing channel
2. Added Qwen3-tts and Qwen3-asr channels
3. Separate settings for F5-TTS/Index-TTS/Spark-TTS/Dia-TTS
4. Optimized processing speed for audio acceleration and video slow-down, using multi-threaded concurrent processing
5. Added audio translation feature; the first button on the left side of the main interface is now "Translate Video and Audio"
6. Refactored interface language settings for easy customization of new language interfaces (Copy the file videotrans/language/zh.json, rename it to the target language code, use AI to translate the value fields, then start the software, select the new target language in advanced options, and restart.)
7. Unified prompt files moved to the videotrans/prompts directory
8. Optimized model download: First check connectivity to huggingface.co; if unavailable, force use of the domestic mirror hf-mirror.com
9. Added Minimaxi dubbing channel
10. Refactored some underlying code for better maintainability
11. Added a Restart button in the bottom right corner
12. Removed the subtitle editing translation function panel and YouTube download feature

If you haven't installed an older version, download the full version here. For CUDA acceleration, you need an NVIDIA GPU and to install CUDA 12.6 and cuDNN 9.

Baidu Netdisk Download: https://pan.baidu.com/s/12DeI5M8zxJcd98G2vkDIBQ?pwd=j7bc

Huggingface.co: https://huggingface.co/mortimerme/repocollect/resolve/main/win-videotrans-v3.81.7z?download=true

v3.81 Patch Package (360MB)

Cannot be used alone; only for overwriting old packages. The difference from the full package is that it does not include the torch module. Since torch was upgraded from 2.5 to 2.7, the patch package does not include torch. Directly downloading and overwriting with the patch may disable CUDA acceleration; it is recommended to download and overwrite with the full package.

Baidu Netdisk Download: https://pan.baidu.com/s/1VQi6vwj7pfOQyzQI1v0yFg?pwd=rm7j

Huggingface.co: https://huggingface.co/mortimerme/repocollect/resolve/main/win-PatchUpdate-3.81.7z?download=true

Extraction Notes:

• Extract the 7z archive. After extraction, find the sp.exe file and double-click to start the software. Do not run it directly from within the archive.
• Do not extract the software to the desktop or folders requiring administrator permissions like Program Files on the C drive.
• Avoid path errors: Do not include Chinese characters, spaces, or special symbols in the extraction path.
• If you encounter errors when extracting with 7zip, try using 360 Compression software (https://yasuo.360.cn).

---

Online Q&A BBS

---

faster-whisper Model Download Tool, Manual Model Download

By default, the software automatically downloads the required models. If downloads fail frequently or you want to download manually, you can download a specialized model download tool here.

1. Download the Tool:

   • For users in China (Recommended):

     • Click to download down_hf-mirror.exe
     • (Note: This version downloads from a domestic mirror site, offering fast speeds and usually requires no additional network setup.)

   • For overseas users or those with stable proxies:

     • Click to download down_huggingface.exe
     • (Note: This version downloads from the Hugging Face official website, requiring stable access to the international internet or a configured system proxy.)

2. Usage Instructions:

   • Place the downloaded .exe file in the same folder as the main program (together with sp.exe).
   • Double-click the downloaded .exe file. It will open a command-line window and automatically start downloading all required models.
   • The download process may take some time, especially since the large series models are very large. Please wait patiently until the window indicates completion.

Important Notes:

• Models like large-v1, large-v2, large-v3 are very large (several GB). If your network is unstable, downloads may fail midway.
• If the download fails, don't worry, simply double-click the same .exe file again. It supports resuming from breakpoints and will continue from where it left off.

---

MacOS/Linux System Source Code Deployment

Full source code repository address: https://github.com/jianchang512/pyvideotrans

• Detailed Tutorial for MacOS System Source Code Deployment
• Detailed Tutorial for Linux System Source Code Deployment
• Deploying This Software via Source Code Using the uv Tool

---

Download Historical Versions