Translation
Document translation
Upload DOCX, PDF, XLSX, PPTX or bilingual XLIFF files, translate their text, and get the result back in the original file type.
After upload, DeepReference extracts the text, splits it into segments, applies the selected settings and language assets, and rebuilds the translated file. Structure, styles, tables, and images are retained where the source format and rebuild process support them.
Step by step with screenshots: Document translation.
Translate with AI or in the translation editor
In the upload window you choose the languages first, then how the document gets translated, then the file. Every option has a small (i) beside it that explains what it does.
| Choice | What happens | Words used |
|---|---|---|
| Translate with AI | Translation memory is checked first (when it is on). Exact matches are reused and everything else is translated by AI, with your glossary terms. You download the file or review it in an editor. | The words the AI translates |
| Open in the translation editor | The document opens in the Translation editor with the source split into segments and the translation side empty. Only exact translation memory matches are filled in. You translate, and click Translate on any segment where you want AI help. | None to open; only the segments you ask AI to translate |
The translation editor route works with Word, Excel, PowerPoint and bilingual XLIFF files from the Starter plan up. PDFs are translated with AI and can then be reviewed in the Translation editor. A document opened this way counts toward your monthly document limit.
Supported formats
| Format | Notes |
|---|---|
| DOCX | Rebuilds supported styles, tables, images, headers and footers, lists, and inline formatting. |
| Requires readable, selectable text, including a searchable OCR layer. Raw image-only scans need OCR before upload. Extracts translatable text and rebuilds a PDF that follows the source layout where possible. | |
| XLSX / XLS | Cell-by-cell translation; formulas and numbers are left alone. |
| PPTX / PPT | Translates slide text using the source presentation structure. Check text fit and layout in the exported file. |
| XLIFF / SDLXLIFF / MQXLIFF | Bilingual files from a CAT tool (Starter and up). Only empty target segments are translated; locked, confirmed and translated segments and inline tags stay as they are. See Working with other CAT tools. |
What happens to your file
Extraction and segmentation
The text is extracted into reviewable segments while the software records supported source structure and formatting for rebuilding.
Translation with your assets
Exact translation memory matches are reused first, from the memories selected for this document. When your glossary outranks memory, a match that leaves out a glossary term is translated again with the term instead. Everything else goes to the AI with your glossary terms. See Translation memory.
Rebuild
Translated text is placed into the saved source structure. Longer target text, floating objects and unavailable fonts can affect layout. Inspect the exported document rather than assuming identical pagination.
Review or download
In the document list, use Open to choose an available editor or Download. PDF text review uses the Translation Editor; the Document Editor is for eligible DOCX jobs. Editors are available on computers and tablets.
Find a document: search and filters
Recent documents on the dashboard and the full Documents page share the same search and filters. Search looks at file names and at the text inside a document, in the source and in the translation. Filters narrow the list by language pair, settings (AI translation, translation editor, terminology only, used translation memory, used a glossary), status (completed, in the editor, processing, draft, failed) and date (preset ranges or your own). Filters combine, and they cover every document in your account, not just the page you are looking at.
A document opened in the translation editor shows as In the editor with its progress, for example "In the editor · 12 of 80 translated".
Without an account and on the Free plan
You can translate a document straight from the Document tab on the home page, without an account: 1 document a month. A free account includes 1 document and 1,000 words a month. In both cases the document is translated with AI and downloaded; glossaries, translation memory, tones and the editors come with a paid plan.
Background processing and email delivery
Processing continues on the server if you close the browser window. Only the Cancel command stops the job. Select email me the result during upload, or enable it in settings, to receive a completion email.
Layout still needs a final check
The rebuild uses structure and formatting captured from the source, but translated text can be longer or shorter and complex PDFs are not always recoverable exactly. Inspect the downloaded file before delivery. For DOCX jobs, use the Document Editor to review translated text in the rendered page layout.
Apply a saved preset for your language pair, translation tone, and glossary. If you also extract terminology, the preset can supply its domain separately. Before a large job, run an Analysis report to check translation memory coverage.
Keep exploring
Translation editor
Review source and target segments side by side with Translation Memory, terminology, QA, comments, status, and per-segment retranslation.
Document editor
Review and edit a translated DOCX in its rendered page layout, with tables, headings, images, and segment tools in view.
Exporting your work
Download the latest saved translation, choose other available formats and inspect the rebuilt document before delivery.
Analysis report
See the source-word total, internal repetitions, translation-memory coverage, and a transparent workload estimate before you begin a job.
Revision comparison
Compare two versions of a document, see every class of change, and if the original was translated here, update only the sentences the revision touched.