Pre-translate Entire Word Documents with DeepL in One Pass

Pre-translating a whole document in one pass is something surprisingly few tools make easy. Most CAT tools either don't offer batch pre-translation at all, or tuck it away behind a fiddly setup.

DeepL2TMX takes the direct route: hand it a Word file and it gives you back a segmented, bilingual TMX and/or XLIFF, pre-translated with DeepL – ready to drop into memoQ as LiveDocs, load into any translation memory, or use however suits your workflow.

DeepL2TMX screenshot

Key Features DeepL2TMX

This small desktop tool is built to facilitate a faster start on a new job. The push to built it was personal: I needed a quick reference translation to load into memoQ LiveDocs for a job I was working on. So I bought a DeepL API key – only to find out I couldn't use it: the new DeepL API key requirements meant my version of memoQ could no longer connect to it, and upgrading wasn't a road I wanted to take just for that.

The key sat there, paid for and worthless.

I tried the trial versions of a few other CAT tools, but they only pre-translated a limited number of segments at a time. So I vibe-coded this small tool of my own: hand it a Word file, and it runs the whole document through DeepL at once and returns a clean TMX and/or XLIFF to use as reference. It turned out to be useful well beyond that first use case, so I'm sharing it here for fellow translators.

DeepL2TMX is:
  • built to keep source and target in clean 1:1 sentence alignment;
  • set up to keep your DeepL API key stored locally on your own machine;
  • able to write a ready-to-use bilingual TMX and XLIFF;
  • compatible with any CAT tool that reads generic TMX and XLIFF 1.2;
  • able to generate a quick segment/word/character-count analysis.

How to Use DeepL2TMX

First, enter your DeepL API key (any DeepL API plan works – the app selects the right endpoint automatically) and pick the monolingual .docx you want to prepare. Optionally, tick Remember to have the app remember your key for next time.

Next, choose your source and target language, optionally set the formality, and select which output you need: TMX, XLIFF, and/or the plain-text analysis.

Pick an output folder and click Translate. DeepL2TMX segments your document into sentences, sends them to DeepL for translation, and writes the aligned files straight to that folder.

Install DeepL2TMX

DeepL2TMX is a single Python script, just download and run it locally.

Requirements

Python 3.9+ and a DeepL API key. Install the dependencies with:
pip install -r requirements.txt

Requires Python 3.9 or higher. pysbd is optional but improves sentence segmentation.

get it on GitHub

About DeepL2TMX

The DeepL2TMX desktop app is developed and maintained by BlacKKite Translations.

Privacy Policy

DeepL2TMX does not collect, store, or transmit any usage data of its own. The only data that leaves your machine is the source text you choose to translate, which is sent to the DeepL API under DeepL's own terms and privacy policy. Your API key and preferences are stored locally in a config file in your home folder — never inside the project folder, and never uploaded anywhere.

Release Notes
  • Version 1.0 – 18-07-2026
    First public version: turns a monolingual .docx into a sentence-aligned, DeepL-translated TMX, XLIFF and analysis, with a manually chosen output folder and generic XLIFF 1.2 output for use in memoQ and other CAT tools.


Author bio

English to Dutch translator/editor Erwin van Wouw

Erwin van Wouw (1973) is a native Dutch technician turned translator based in Utrecht, the Netherlands.

With over 18 years of experience as a professional Dutch translator and proofreader, he assists businesses in transforming their texts into impeccably written and smoothly flowing Dutch documents. Erwin specializes in translation services for the engineering, sustainable technology, printing equipment, consumer electronics, and IT industries.