How to Convert a Scanned PDF to an Editable Word File
By Tooldesk team · · 4 min read
To turn a scanned PDF into an editable Word file, you need OCR (optical character recognition), which reads the letters in the page images. Open our PDF to Word converter, keep Recognise text in scanned pages (OCR) switched on, add the PDF and select Convert to Word. Scanned pages are straightened, read, and rebuilt as paragraphs and tables in a .docx file. Everything runs in your browser, so the document is not uploaded.
How to tell if your PDF is scanned
Open the PDF and try to select a word with your cursor.
- You can highlight single words: the PDF has real text. Any converter can read it, and OCR isn’t needed.
- The cursor draws a box over the whole page: the page is an image. It came from a scanner, a phone camera app or a “print to image” step, and it needs OCR.
Many PDFs mix both, for example a typed report with a scanned signature page. Our converter checks each page and only runs OCR where there is no text.
Steps to convert a scanned PDF
- Open PDF to Word.
- Add the scanned PDF.
- Make sure Recognise text in scanned pages (OCR) is ticked.
- Select Convert to Word. The first time, the text recognition engine downloads once (a few MB) and is then cached.
- Wait while each page is read. A clear page usually takes several seconds. Pages with tables take longer because each cell is read on its own.
- Download the
.docxand proofread names, numbers and dates.
What OCR does to your page
Good OCR is mostly image clean-up before the reading starts. Our converter:
- Straightens tilted scans. A page fed into a scanner at a slight angle is rotated back to level. This matters more than it sounds: table borders only line up on a straight page.
- Finds ruled tables. It detects the horizontal and vertical lines, works out the grid, and reads every cell separately. The result is a real Word table instead of a jumble of numbers.
- Turns pages the right way up. A page scanned sideways or upside down is detected and rotated.
- Boosts contrast. Faint, grey scans are normalised so letters stand out from the paper.
How to get the most accurate text
OCR accuracy depends far more on the scan than on the software. These settings make the biggest difference:
| Setting | Use | Avoid |
|---|---|---|
| Resolution | 300 DPI for normal text, 400 for small print | 72–150 DPI |
| Colour | Greyscale or colour | Harsh black-and-white on faint originals |
| Angle | Page flat on the glass | Curled pages, book spines |
| Phone photos | Even daylight, camera straight above | Shadows, flash glare |
| File | The original scan | A scan that was compressed to a tiny size |
If you scanned with a phone, use the “document” mode of your scanner app. It crops and flattens the page before saving.
Don’t compress a scanned PDF before converting it. Strong compression blurs letters, and blurred letters are misread. Compress the PDF after you have what you need.
Other ways to OCR a scan
Google Docs. Upload the PDF to Google Drive, then right-click it and choose Open with → Google Docs. Docs runs OCR and puts the text under an image of the page. It is free and accurate, but tables and most formatting are lost.
Microsoft Word. Word converts PDFs with real text, but it has no OCR of its own. A scanned page opens as a picture.
Single images. If you have a photo or screenshot rather than a PDF, use Image to Text instead. It runs the same clean-up and supports more languages, including Urdu, Hindi and Arabic.
Checking the result
OCR is never perfect. Proofread anything that matters, especially:
- Numbers: 0 and O, 1 and l, 5 and S are the classic swaps.
- Names and email addresses, which aren’t in any dictionary.
- Totals in tables. Compare a few against the original.
For tips on improving difficult images, read why OCR gets text wrong. For layout problems on PDFs that already have text, see our guide to keeping formatting from PDF to Word.
Questions people ask
Can I edit a scanned PDF without converting it?
Not the text itself. A scanned page is an image, so you can draw on it or add text boxes, but you can’t change the words until OCR has turned them into text.
Is OCR accurate enough for legal or financial documents?
On a clean 300 DPI scan, most words come out right, but a single wrong digit can matter. Always check figures against the original before you rely on them.
Why is my converted document full of nonsense characters?
Usually the scan is too low in resolution, too faint, or heavily compressed. Rescan at 300 DPI in greyscale and convert again.
Does OCR work for handwriting?
Only roughly. OCR engines are trained on printed text, so neat block capitals may work and joined-up handwriting usually won’t.