Translate PDF Online - Keep the Layout
Translate PDF files between 47 languages right in your browser. Get the same-layout PDF, an editable Word file or a bilingual version. No upload, no sign-up.
- Free, no sign-up
- Files stay on your device
- No watermark
By using this tool you agree to our Terms and Privacy Policy.
Large files can take a little while. Keep this tab open.
Your file is ready
Something went wrong
Translate PDF in 3 steps
Choose a PDF with selectable text (scans need OCR PDF first).
Check the detected language, pick the target language and the outputs you want.
Click Translate and download the translated PDF or Word file. Nothing is uploaded.
Private PDF translation that keeps your pages looking the same
Same layout, translated
Each paragraph is translated and placed back in its own box on the original page. Images, tables, colours, links and anything that is not text stay exactly as they were, and the cover behind the new text uses the background colour sampled from the page.
PDF, Word or bilingual
Download the translated PDF, an editable Word document in reading order, or a bilingual Word file with the original and the translation side by side, paragraph by paragraph. Pick one, two or all three.
Translated on your device
An open-source AI translation model runs inside your browser. Your document is never uploaded, there is no account and no watermark. The model downloads once (about 115 MB for most language pairs) and is reused from your browser cache.
47 languages, detected for you
The document language is detected automatically. Common pairs such as English with Romanian, German, French, Spanish or Italian use small dedicated models; other pairs go through English or a universal 100-language model. Translate only the pages you need.
How to translate PDF files and keep the layout
Many online translators send your document to a server and return plain text that loses the original look. BestConverter translates the PDF on your own device and writes the translation back onto the original pages. Contracts, reports, manuals and school papers stay private because they never leave your computer or phone.
What happens to your file
- The text of each page is read together with its position, size and colour.
- Lines are grouped into paragraphs, headings, list items and table cells, and the language is detected.
- Each sentence is translated by an open-source neural model running in your browser. Repeated headers and footers are translated once.
- The original lines are covered with their own background colour and the translated paragraph is set in the same box, shrinking slightly when the translation is longer.
Tips for the best result
- Check that the detected source language is right. Choosing it yourself helps with short or mixed documents.
- Use the Pages field to translate just the part you need. Long documents finish much sooner, and the pages you skip stay unchanged in the PDF.
- Need to edit the wording? Choose the Word output, or open the translated PDF in the editor from the result window.
- For learning a language or checking a translation, the bilingual Word file shows each original paragraph next to its translation.
- On a phone, keep the Fast model and translate fewer pages at a time to avoid running out of memory.
Limits to know
Password-protected PDFs open after you type the password, but an encrypted file cannot be rewritten as a same-layout PDF: unlock it with Unlock PDF first, or take the Word output. Machine translation is a fast way to understand a document, not a certified translation. For official use, have the result reviewed by a qualified translator.
More PDF tools
Frequently asked questions
Is it safe to translate a PDF online here?
Yes. The text is read and translated inside your browser, so the PDF is never uploaded. The only download is the translation model itself, from huggingface.co, the first time you use a language pair; after that it is cached and the next translation in that pair needs no download at all.
How accurate is the PDF translation?
It uses OPUS-MT (University of Helsinki) and M2M-100, open-source neural translation models. For common pairs such as English to German, French, Spanish, Italian or Romanian, results are usually clear and readable. Pairs routed through English and the universal model are less polished, so have important legal or medical documents checked by a person.
Can I translate a PDF and keep the formatting?
Mostly, yes. Text is replaced paragraph by paragraph in the same position, size and colour, and may be set a little smaller when the translation is longer. A standard sans-serif font replaces the original typeface, and the original text is covered rather than deleted. If a paragraph does not fit, the Word output has the full text.
Why does the first translation take so long?
The first time you translate between two languages your browser downloads the model: about 115 MB for most pairs, about 230 MB for pairs translated through English and about 650 MB for the universal model. Translation then runs on your own processor, so a long document can take several minutes. Progress, a time estimate and a Cancel button are shown.
Can I translate a scanned PDF?
A scan is a picture of text, so there is nothing to translate yet. Run it through OCR PDF first to make the text readable, then translate the result here.
Can I translate a PDF into Chinese, Arabic or Japanese?
Yes, but as a Word document. The same-layout PDF works for target languages written in Latin, Cyrillic or Greek letters. For Arabic, Hebrew, Persian, Hindi, Bengali, Thai, Vietnamese, Chinese, Japanese and Korean you get a Word file, which displays those scripts correctly.