Converting bilingual translation files to the MT format: file.source and file.target

Converts a TMX translation memory into two files, source and target, for machine translation training.

There are two scripts that do the same thing, one with XML minidom and one with XML ElementTree. Please note that TMX2MT-ElementTree.py is supposed to be faster, and supports segments that contain inline tags.

Whatever script you decide to use, run it in the Terminal/CMD as follows:

python3 <script.py> <tmx_file_name.tmx> <source_lang> <target_lang>

• Check out the new notebook (here) that uses the xmltodict library. The notebook converts a sophisticated TMX translation memory into two formats:
  • Moses format (text file, one segment per line)
  • GPT fine-tuning format (JSON lines / JSONL)

If you have questions or suggestions, please feel free to contact me.