AIVideo Translator is an open-source project for automated video translation using artificial intelligence. It combines speech-to-text, machine translation, and text-to-speech technologies to efficiently transfer spoken content into other languages.
I started this project to translate the great #TYPO3 traning videos which are produced and provided by Wolfgang Wagner, but were only available in German. As we in web-vision are a team of people with multiple cultures and languages, we had the need to translate those videos to English.
Big Kudos to Wolfgang Wagner for providing us with the videos, which we now can use for internal trainings of TYPO3 beginners.
To check the quality of the AIVideo Translator you might want to check these short clips of Wolfgang Wagner talking about Search in TYPO3 on YouTube:
- Transcription of spoken video audio
- Translation into multiple languages
- Output as a subtitled file or synthetic audio track
- Support for modern AI models (e.g., OpenAI Whisper, Transformers)
- Customizable dictionaries
- Operating System: Linux, macOS, or Windows 10+
- Python: Version 3.9 or higher
- RAM: Minimum 8 GB (recommended: 16 GB for large videos)
- GPU: Optional (for faster processing via CUDA)
- ffmpeg: Must be installed and available in the system PATH
- pip: Python Package Installer
-
Clone the repository:
git clone https://github.com/web-vision/aivideotranslator.git cd aivideotranslator -
Create a virtual environment (optional, recommended):
python -m venv venv source venv/bin/activate # Windows: venv\Scripts\activate
-
Install dependencies:
setup.sh
-
Run maim menu:
python main_menu.py
- You can use the extracted SRT-Files for providing sub-titles in various languages. Also, it's recommended to check them occasionally as sometimes special terms and phrase are not well translated. Correct them with any text editor you want.
- Use the glossary.py file to define various notations or even wrong spellings of brands or terms and phrases. Some exmaples for "TYPO3" are already included.
- Check the languages selection first. Some speakers sound more natural than others. We use "p230" which sounds like a native English speaker to us (at least for my German ears).
- You can configure some things, as the speaker or the type of subtitles you want to have in the video_congfig.py
This project uses the following open-source libraries:
- OpenAI Whisper – Speech-to-Text (MIT License)
- transformers – Machine Translation (Apache 2.0)
- ffmpeg-python – Video and audio processing (MIT License)
- gTTS – Google Text-to-Speech (MIT License)
All listed components are compatible with GPLv3.
This project is provided without any warranty. Use at your own risk. The author assumes no liability for any damages, data loss, or other consequences resulting from the use of this software.
This project is licensed under the GNU General Public License Version 3 (GPLv3).
For more information, see the LICENSE file or visit https://www.gnu.org/licenses/gpl-3.0.en.html.
Pull requests are welcome! For major changes, please open an issue first to discuss what you would like to change.
For questions or suggestions, feel free to reach out at [b.hinzer@web-vision.de].