To convert scanned PDF to Word, you first need OCR, because a scan is only a picture of a page and contains no real letters. Run the file through the PDFNova OCR PDF tool to recognise the text, then move that text into Word, either by pasting it or by converting the new searchable PDF. The whole job is free and happens in your browser.
The quality of the result depends far more on the scan than on the software, so a few minutes of preparation pays off.
Why a scanned PDF will not convert directly
A normal PDF exported from a word processor contains text as characters. A converter can read those characters and build a document from them.
A scanned PDF is different. Each page is a photograph. To a computer, the words are just dark shapes on a light background. If you put a scan straight into a PDF to Word converter, you get an empty or nearly empty file, because there is nothing to extract.
OCR, short for optical character recognition, looks at those shapes and works out which letters they are. Once that is done, you have real text to work with.
Not sure which type you have? Try to select a sentence in the PDF. If words highlight one by one, it is text-based and you can go straight to PDF to Word. If nothing highlights, it is a scan and you need the steps below.
How to convert a scanned PDF to Word, step by step
- Open OCR PDF and add your scanned PDF or image.
- Choose the language of the document. The tool supports English, Urdu, Arabic, Hindi, French, Spanish, German and Chinese (Simplified).
- Start the recognition. The first time you use a language, its language data downloads, so allow a little extra time and use a stable connection.
- When the text appears, read through it quickly to judge the accuracy.
- Choose your output: copy the text, download a .txt file, or download a searchable PDF.
- Move the text into Word using one of the two routes described next.
- Proofread the Word file against the original scan.
Recognition runs on your device. PDFNova processes files in your browser, so your documents are not uploaded to a server.
Two routes into Word, and which to pick
| Route | How it works | Best when |
|---|---|---|
| Copy and paste | Copy the recognised text and paste it into a new Word document, then apply your own headings and styles | You want full control, the document is short, or you are using your own template |
| Searchable PDF, then convert | Download the searchable PDF from OCR, then open it in PDF to Word to get a .docx with headings and paragraphs detected | The document is long and has a simple, one-column layout |
The copy and paste route is the most predictable. You get exactly the text you saw on screen, and you format it once. If you only need the words and not a Word file, our guide on how to convert a PDF to plain text covers that case.
With the second route, remember that the text layer comes from recognition, so any OCR mistakes carry over into the .docx. Images and complex layouts are not reproduced exactly either. Expect to tidy the result.
Get a better scan before you run OCR
OCR accuracy depends on scan quality. A clean scan can be recognised with few errors. A dark, tilted phone photo produces nonsense.
- Use good, even light. Avoid shadows from your hand or phone.
- Keep the page flat. Curved pages near the spine of a book distort the letters.
- Shoot straight on. An angled photo stretches the text.
- Fill the frame. The page should take up most of the image so the letters are large enough.
- Pick a high-contrast look. Black text on a white background reads best.
If you still have the paper, it is often quicker to scan it again properly than to correct a bad result. The Camera Scanner detects page edges, corrects perspective and offers a black and white filter, which gives OCR a much easier image to read.
Fixing common OCR errors in Word
Similar-looking characters are mixed up
OCR often confuses the letter l with the digit 1, the letter O with zero, and rn with m. Check every number, amount, date and reference code by eye. These matter most and spell-check will not catch them.
Lines break in the middle of sentences
Recognised text usually keeps the line endings of the scan. Remove the extra breaks so each paragraph flows naturally, then apply a paragraph style.
Tables come out as jumbled text
OCR reads across the page and does not understand cells. Rebuild the table in Word and type the figures in while looking at the scan.
Urdu or Arabic text has many mistakes
Choose the correct language before recognising. Printed Arabic and printed English are easier to read than Urdu in Nastaliq script, where letters join and stack in ways that are hard for software. Expect more manual correction for Urdu, and far more for handwriting of any kind.
Words from the next column appear mid-sentence
Multi-column pages can be read straight across. Sort the text into the right order by hand, one column at a time.
Other things you can do with the recognised text
Word is not the only destination. If your real goal is to find words inside the scan, keep it as a PDF and read how to make a scanned PDF searchable. The page still looks like the original, but you can search and copy from it.
If you downloaded the .txt file and later want a neat, paginated document from it, see how to convert a TXT file to PDF.
Frequently asked questions
Can I convert a scanned PDF to Word on my phone?
Yes. The OCR tool works in a modern phone browser. Long documents take more time on a phone, so be patient and keep the tab open while it works.
How accurate is OCR on a scanned PDF?
It depends on the scan. Sharp, straight, high-contrast pages in a printed font come out well. Faint photocopies, stamps over text, handwriting and decorative fonts reduce accuracy.
Will the Word file look like the scanned page?
No. You get the words, not the design. Logos, signatures, stamps and exact positions are not carried over, so rebuild any layout you need in Word.
Does OCR work on a photo of a document, not just a PDF?
Yes. The tool recognises text in scanned PDFs and in images, so a clear photo of a page works too.
Scan cleanly, choose the right language, then proofread the numbers and names. Start with the free OCR PDF tool to turn your scan into text you can edit.