Translate a PDF
Page by page, and it remembers where you stopped. Scanned pages get read too. How it works →
Translating a PDF, including scanned ones
A PDF is either text with a layout or a picture of a page, and the two need different handling. Traxlate reads which one it has: where there is a real text layer it is translated and laid back into the page, and where the page is a scan the text is recognised on the device first.
The result keeps the page structure rather than becoming a wall of paragraphs, so a contract still reads as a contract and a form still looks like a form. You can review the result page by page before exporting it.
When the text layer is broken
Some PDFs carry a text layer that looks fine to a program and is not real spelling. Persian and Arabic files often store presentation forms, so copying the text out gives letters in the wrong shapes. Traxlate detects that case and reads the page as a picture instead of trusting the broken layer, because a confident wrong extraction is worse than none.
Why it runs on your device
The engine and the language pack download once and then run locally, for all 66 languages. Nothing you translate is sent anywhere — not the easy sentences and not the hard ones, because the server path was removed in August 2026 and there is nothing behind this page to send them to. You can check that by turning the network off and translating anyway. The privacy model says exactly what does and does not leave the machine.
Related tools
- Translate Word and other documents — formatting kept
- Read text from a picture — signs, labels, screenshots
- Review before delivering
Common questions
- Can it translate a scanned PDF?
- Yes. Pages without a usable text layer are recognised on your device first, then translated. The recognizer downloads its language data once, a few megabytes.
- Is my PDF uploaded to a server?
- No. Both the reading and the translation happen on your machine.
- Does the layout survive?
- The page structure is kept and the translated text is laid back into it. Text that grows in translation is fitted to its box, and you can review each page before exporting.
- Can I edit the translation before exporting?
- Yes. The viewer lets you correct any block, and corrections you make are remembered for the rest of the document.
- What about a PDF in a language I did not choose?
- The language can be detected rather than named, including for files whose text layer is damaged.