TransAIO is a lightweight desktop application that automates and optimizes the translations by combining Machine Translation and/or LLM with Translation Memory and Termbase. It is designed with efficiency and simplicity in mind, offering multithreaded performance, contextual translation support, and a user-friendly PyQt6 interface.
- Context-Aware Translation — Uses Translation Memory and Termbase (only with LLMs) for higher-quality results.
- Live Translation Memory — Each translation uses a seperate (temporary) Translation Memory together with the user selected Translation Memory for better translation consistency.
- Fuzzy Match Support - Segments that have 80%+ similarity are considered Fuzzy Match and sent to LLM for better translation consistency.
- Multithreaded Translation Engine — Translates large files quickly using concurrent threads.
- Tag-Safe Translation — Preserves inline formatting tags throughout the process.
- MQXLIFF Compatible — Handles
.mqxlifffiles for seamless integration with memoQ. - Translation Logging — Records each translation operation for audit and review (under _temp folder).
For Windows users, a pre-built executable is available. Download the latest release from the Releases page and run the program directly without needing to install Python or dependencies.
- Python 3.13 or later
Clone the repository:
git clone https://github.com/sbkocoglu/trans-aio.git
cd sourceInstall dependencies (venv recommended):
pip install -r requirements.txtRun the application:
python trans_aio.py- Open Settings menu from top-menu.
- Enter your API keys and/or Ollama host URL and select the Ollama model you have installed.
- Select how you want to translate and improve translations from Translation memory.
- Select an MQXLIFF file.
- (Optional) Select a Translation Memory and/or Termbase.
- Click Start and select a destination to save the MQXLIFF file.
- Wait for translation to end.
- (Optional) Check _temp folder for details.
- OpenAI API: The translation is slow due to API limitations. Your API limitation is based on your OpenAI tier.
- Some rare tag formats may not be restored perfectly.
- If the system has limited threads, heavy jobs may cause UI freezing.
- Only .mqxliff and .mqxlz files are currently supported.
- Improve error handling
- Machine Translation support for other MT engines.
- Additional support for other LLMs
- Further translation improvement (fuzzy matches where only the numbers are different, etc.)
AGPL-3.0 license — see LICENSE for details.
Developed by Serkan Burak Koçoğlu