Installation Instructions
If you are using Windows 10/11, download the complete package, extract it to a path without spaces or Chinese characters, and double-click sp.exe to use it (you must extract it before using it, you cannot directly double-click sp.exe in the compressed package).
Installation tutorial for Linux / Installation tutorial for MacOS
MacOS/Linux upgrade: Re-pull the source code to overwrite, and then execute
pip3 install --upgrade openai-whisper elevenlabspip3 install --no-deps --force-reinstall "faster-whisper @ https://github.com/SYSTRAN/faster-whisper/archive/refs/heads/master.tar.gz"
Change
- Fix: Add a pause wait after each dubbing of the edge-tts channel to avoid 403 errors as much as possible. Fill in a value greater than 5 in Menu-Tools-Options-Advanced Options-Pause after Dubbing.
- Fix: When faster-whisper/GeminiAI is used as a speech recognition channel, the subtitle duration recognized by speech can be adjusted through Menu-Tools-Advanced Options-faster/openai--Minimum Speech Duration/Maximum Speech Time.
Win v3.64 Complete Package Download
If you have not installed the old version, please download the complete version here. If you need CUDA acceleration, you need to have an NVIDIA graphics card and install cuda12.x and cudnn9.
Baidu Netdisk download address (including tiny/medium model): https://pan.baidu.com/s/10UHeNV20I7csrQSybYdPYw?pwd=eei8
GitHub address: https://github.com/jianchang512/pyvideotrans/releases/download/v3.64/win-videotrans-v3.64-tiny.7z
v3.65 Patch Package 190MB
If you have installed the 3.x version, you can download the patch package and extract it to the directory where sp.exe is located to overwrite the existing sp.exe and folder.
Baidu Netdisk download address: https://pan.baidu.com/s/1AFcVVL9n5TfpgiQPfOMPiw?pwd=817i
GitHub address: https://github.com/jianchang512/pyvideotrans/releases/download/v3.64/win-PatchUpdate-3.65.7z
If the prompt
Need to download the complete packageappears after the patch is overwritten, please upgrade the graphics card driver, upgrade cuda to 12.x, upgrade cudnn to cudnn9, and download the complete package again.
All Model Download Addresses
In order to avoid the compressed package being too large, the pre-packaged version only has the smallest model tiny built-in, and the recognition effect is not good. For better models, please click download. It is recommended to use at least the medium model, and large-v2 is recommended
https://github.com/jianchang512/stt/releases/tag/0.0
ffmpeg.exe Download
The Windows pre-packaged version has ffmpeg built-in, so there is no need to download it separately. If it is source code deployment, you need to download the following 2 files to the ffmpeg folder in the software root directory
https://github.com/jianchang512/stt/releases/download/0.0/ffmpeg.exe
https://github.com/jianchang512/stt/releases/download/0.0/ffprobe.exe