How to convert a scanned PDF to Word (free, and nothing is uploaded)

Sist oppdatert: September 16, 2026

A scanned PDF is a stack of photographs. You can see the words, but your computer cannot: there is no text inside the file, only pixels. That is why copying from it gives you nothing, and why a straight conversion to Word produces an empty document. What you need first is text recognition, usually called OCR.

Convert it in your browser

1. Open PDF to Word in the language of your document. The Spanish page reads Spanish scans, the German page reads German ones, and so on. This matters more than anything else below.
2. Choose your scanned PDF.
3. Press Convert to Word. The first scanned page downloads the recognition engine once, a few megabytes, and then reads the pages one by one.
4. The .docx downloads on its own. Open it in Word, Google Docs or LibreOffice.

Recognition is slow compared with everything else on this site, because it is your own device doing the work: think seconds per page, not per document. A long scan is a good moment to leave the tab open and go for a coffee. Nothing is uploaded while you wait — you can check that yourself.

Why the language of the page matters

Recognition models are trained per language. An English model has never seen "ñ", "ø" or "ł", so it replaces them with whatever looks closest: "española" becomes "espafiola", "número" becomes "nimero". Opening the tool on the page for your language loads the right model and those mistakes disappear. If your document mixes two languages, convert it in the one that dominates the body text.

What comes out, and what does not

• The text, rebuilt into paragraphs in reading order, with sentences joined across page breaks and repeated headers, footers and page numbers dropped.
• Pictures, placed where they appear.
• Tables, when their cells are drawn with lines. On a scan the lines are part of the image, so a scanned table usually comes out as text rather than a table.
• Not the exact layout. A converted scan is a document you can edit, not a facsimile.

Read the result before you trust it

Recognition is never perfect. Give the document one pass looking at the things that matter: figures, names, identity numbers, dates. A "5" read as "S" in an invoice total is the kind of mistake that costs money later. Accented capitals and text over grey backgrounds are the usual weak spots.

Getting a better scan in the first place

• Scan at 300 dpi. Below 200 the letters break up; above 400 you only get a bigger file.
• Keep the page straight. A few degrees of rotation cost more accuracy than poor lighting.
• Scan in black and white or grey for plain text, and avoid shadows if you are using a phone. If you are photographing a document, these tips for scanning with a phone apply.

If you only need to search or copy, not edit

There is a shorter route: OCR PDF keeps the scan exactly as it looks and adds an invisible layer of recognised text on top, so the document becomes searchable and you can copy from it while it still looks like the original. That is explained in how to make a scanned PDF searchable. Use PDF to Word when you need to rewrite the content, and OCR PDF when you need to find things in it.

Why do it in the browser

Scans are usually the most sensitive documents people have: contracts, payslips, ID cards, medical letters. Every step here happens on your own device, so the file is never uploaded, never queued on someone's server and never waiting to be deleted in an hour. If you want proof, load the page, turn off your Wi-Fi and convert anyway — it still works.

Se også: Derfor blir layouten feil når du konverterer PDF til Word (og hva som hjelper)