Getting started

Working with other CAT tools

Translate bilingual files from the CAT tool you use today, bring in your translation memories and termbases, and take your work back out whenever you want.

DeepReference is built to sit next to the tools translators already use, not to lock you in. A bilingual file from your CAT tool can be translated here and opened back in that tool. Memories and termbases come in through standard exchange files, and your translation memory goes back out as TMX at any time.

Translating a bilingual file from your CAT tool

Export the project file as a bilingual file, upload it like any document, and download the same file with the empty target segments filled in. Open it in your CAT tool again to review and deliver as usual.

• Supported bilingual files: XLIFF, SDLXLIFF and MQXLIFF, exported from your CAT tool with the target language you need.
• The target language chosen on upload must match the file; a file prepared for another target language is refused. With the source set to auto, the source language is read from the file.
• AI translation and the Translation editor both work. Your glossary applies as with any document; with Translator Pro, exact translation memory matches are filled in first.
• The download is the same file with only the new targets and their segment status changed. Its name starts with DR_ and the language code; rename it if your tool expects the original name.

• Only empty segments are translated. Locked, confirmed, approved and already translated segments, and units marked as not translatable, are left exactly as they are.
• Inline tags stay in place. Formatting and placeholder tags are kept in the target where the translation puts them. If a translation leaves a tag out, the tag is added at the end of the segment so your CAT tool still finds it; check those in the tool's tag QA.
• Status shows what needs review. New targets are marked for review (needs review, Draft, or edited, depending on the format). A segment you approve in the Translation editor is marked translated. Nothing is marked final or signed off.
• The text is the model's own output. DeepReference does not post-edit it. A segment the model could not translate stays empty.

What you can import

Importing memories and termbases

• Memories: TMX, or .sdltm and .mqxlz files directly. Termbases: TBX or CSV.
• Import is on the glossary, termbase or translation memory tab of Language assets.
• Source and target language are read from the file when it contains them; otherwise they are chosen before importing.
• Imported memory segments appear as matches in the Translation editor. Imported glossary terms are sent to the model with the sentences that contain them, and the quality check flags a segment where a required term is missing.

Which plan you need

• Bilingual files (XLIFF, SDLXLIFF, MQXLIFF): Starter and up.
• Glossary imports: Starter and up.
• Termbase (terminology) imports: Freelance and up.
• Translation memory imports and exports: Translator Pro and up.

Taking your work back out

• Translation memory: export as TMX, filtered by language pair or domain if you like.
• Glossaries: export as CSV, which CAT tools can import.
• Approved extracted terms: export as CSV.
• Translated documents: download in the format they arrived in (DOCX, PDF, PPTX or XLSX).
• Bilingual files: download the same XLIFF, SDLXLIFF or MQXLIFF file with the targets filled in, ready to open in your CAT tool.

If you do not use a CAT tool

You do not need one. Upload a Word file and download it translated, with its layout. When a revised version arrives, use Revision comparison to see what changed and retranslate only that part.

Connecting your own systems

Business and Enterprise plans include a REST API with API keys. See the API overview.