OCR PDF
Turn a scanned PDF into searchable, selectable text — choose which pages and which language, right in your browser.
Drop a scanned PDF here, or click to browse
How OCR on a scanned PDF works
OCR (Optical Character Recognition) is what turns a scanned PDF into a normal, working one. A scanned PDF — whether it came from a flatbed scanner, a phone camera, a fax, or an old export that rasterized every page — is really just a picture of a page. There's no real text underneath it, which is why you can't press Ctrl+F to find a word, can't select a sentence to copy it, and a screen reader has nothing to read. OCR reads each page image, figures out what the characters say, and this tool then lays that recognized text invisibly on top of the original page image — so what you see stays exactly the same, but the page itself becomes searchable, selectable, and copyable.
Everything happens on your device using a WebAssembly build of the open-source Tesseract engine — nothing is uploaded anywhere, so your document never leaves your browser tab. That also means recognition speed depends on your own device rather than a server queue: a modern laptop will usually get through a page in a few seconds, while an older phone will take longer. You can pick exactly which pages need OCR (skip the ones that already have real text), choose up to three languages for documents that mix languages on the same page, and turn on Faster mode to trade a little accuracy for noticeably quicker recognition on long documents.
Common use cases
- Making an old scanned contract, form, or legal document searchable before filing it
- Pulling text out of a photographed receipt, ID, or handwritten-adjacent printed note
- Preparing scanned textbook pages, research papers, or class notes for keyword search
- Extracting quotes or data from a scanned report without retyping it by hand
- Making a bilingual or multi-language scanned document (e.g. English + Hindi) properly searchable
- Adding a text layer so a screen reader can read a document that was originally just a scan
For best results, use clean, well-lit, reasonably straight scans in the language(s) actually present in the document — very skewed, blurry, low-contrast, or handwritten pages will recognize less accurately, and selecting a language that isn't in the document tends to hurt accuracy rather than help it. If a page still doesn't contain any selectable text after running OCR, try a sharper rescan of just that page before rerunning.
Related tools
OCR PDF — frequently asked questions
It depends heavily on scan quality. Clean, well-lit, straight scans of printed text typically recognize very well. Handwriting, blurry photos, low-contrast scans, or heavily skewed pages will have more mistakes — always check the extracted text before relying on it, especially for anything you plan to search or quote from later.
Roughly a few seconds to about a minute per page on a typical laptop, since everything runs on your device rather than a server. Two things make it faster: OCR only the pages that actually need it (skip the pages that already have real text), and turn on Faster mode, which recognizes each page at a lower resolution — quicker, at a small cost to accuracy on tiny text.
No. The PDF, the recognition itself, and the output all stay on your device the entire time — nothing is sent to a server.
It looks pixel-for-pixel identical to your original scanned pages, but an invisible, selectable text layer is added underneath. That's what lets Ctrl+F find a word, lets you highlight and copy a sentence, and lets a screen reader read the page aloud — none of which work on a plain scan.
Yes — select up to three languages before running OCR and the recognizer looks for all of them at once, which noticeably improves accuracy on documents that switch between, say, English and Hindi on the same page. Selecting languages that aren't actually in the document tends to hurt accuracy rather than help it, so only pick what's really there.
Yes. After uploading, every page shows as a thumbnail you can tap to include or exclude — useful when only a handful of pages in a longer PDF are actually scans, since OCR is the slowest step in the tool and skipping pages that don't need it saves real time.
If your PDF needs a password to open, this tool will tell you upfront and point you to the Unlock PDF tool first — OCR can't read an encrypted file's pages. If an individual page turns out to be corrupted or unreadable mid-run, the tool skips just that page, keeps going, and tells you exactly which page(s) it couldn't process instead of failing the whole job.
Searchable PDF if you want to keep the original look and just add search/copy ability — this is the right choice most of the time. Plain text if you only need the words themselves (for pasting elsewhere, or a quick read) and want to skip the extra step of rebuilding a PDF, which is a little faster.