Changelog

• Feat: Support new openai models gpt-4o-mini-tts / gpt-4o-transcribe
• Fix: Optimize hardware acceleration during video processing
• Fix: Upgrade F5-TTS interface Use official interface

Windows 10/11

1. Click to enter the pre-packaged version page Select the GitHub address or Baidu Netdisk address, and download the complete package (7z compressed package).
2. Extract the 7z compressed package. After extracting, find the sp.exe file and double-click it to start the software.
3. Extraction Notes
   • Avoid Permission Issues: Do not extract the software to the desktop or folders that require administrator privileges, such as Program Files on the C drive.
   • Avoid Path Errors: The extraction path should not contain Chinese characters, spaces, or special symbols.

MacOS/Linux Systems

Source code address: https://github.com/jianchang512/pyvideotrans

0. First, install conda
1. Create a virtual environment using python3.10, execute the command conda create -n videotrans python=3.10. If prompted, enter yes and press Enter.
2. Activate the virtual environment, execute the command conda activate videotrans
3. Create an empty folder to deploy the source code. Assuming /data/pyvideo has been created, enter the folder and pull the source code from GitHub by executing the command git clone https://github.com/jianchang512/pyvideotrans .
4. Install dependencies by executing the command pip install -r requirements.txt and wait for completion.
5. Install ffmpeg. Execute yum install ffmpeg on CentOS, apt-get install ffmpeg on Ubuntu, and brew install ffmpeg on MacOS.
6. If there are no errors, execute python3 sp.py to open the software.

---

All Models

To avoid an excessively large compressed package size, the pre-packaged version only includes the smallest model, tiny, which has poor recognition performance. It is recommended to use at least the medium model, and large-v2 is recommended.

View all model descriptions

Historical versions