OCR a PDF Online for Free
Make a scanned PDF searchable. Text is recognised in your browser and added back as a real, invisible text layer — the page looks unchanged.
Drag & drop a PDF file here
or browse
Picking the right language markedly improves accuracy. The first run in a new language downloads its recognition model, which takes a moment; after that it is cached by your browser.
OCR PDF makes a scanned document searchable and selectable without changing how it looks. It recognises the text in each page image and adds it back as an invisible text layer sitting exactly on top of the original — so the page still looks like the scan, but you can now search, select, and copy text from it, or open it in the built-in editor and click directly into existing paragraphs. Recognition runs locally, using the correct language model for accuracy.
What you can do
- Adds an invisible text layer so a scan becomes searchable and selectable while looking exactly as before.
- Pick the document's Language for better accuracy.
- Download the searchable PDF, or just the recognised text as Word (.docx) or plain text (.txt).
- Open the result in the editor and click straight into existing paragraphs.
How it works
- 1Drop your scanned PDF onto the page.
- 2Choose the Language of the document — accuracy depends heavily on picking the right one.
- 3Click Process File and wait while each page is recognised (the first run in a new language downloads its model once).
- 4Download the searchable PDF, or use the Word (.docx) or Text (.txt) buttons to save just the recognised text.
Why use SmilingPDF for this
- Text recognition runs in your browser via a WebAssembly engine — the scanned pages are never uploaded to a server to be read.
- The output keeps the original page images untouched, just with a searchable text layer added underneath.
- Recognised text can also be exported straight to Word or plain text, not just a searchable PDF.
Private by design
Your file stays on your device.
- No upload
- No server storage
- No account required
- Browser-based processing
- No watermark
This tool runs inside your browser tab. Your file is read from your device, processed there, and handed back as a download — it is never sent to a SmilingPDF server, so there is nothing for us to store, read, or lose. You can check for yourself: open your browser's developer tools, switch to the Network tab and run the tool. Like most sites, SmilingPDF uses Google Analytics to count visits and which tools are used; that never includes the contents of your files. The privacy policy lists everything the site does collect.
More tools you might like
All free. All private. All in your browser.
AI PDF Summarizer
Get a PDF's key points, summarized on your device
PDF to Text
Extract the text from a PDF as a plain .txt file
PDF Word Counter
Count words, characters and pages inside a PDF
Extract Images
Pull every picture out of a PDF into a ZIP
Invoice to Excel
Upload multiple invoices, get one spreadsheet row each
Merge PDF
Combine multiple PDFs into one document
Compress PDF
Reduce file size without losing quality
PDF to Word
Convert PDF to an editable Word document
FAQ
Frequently asked questions
Got questions? We've got answers.
What does OCR do to my PDF?
OCR (optical character recognition) reads the words in a scanned page image and adds them back to the PDF as real, invisible text. The page still looks exactly the same, but you can now search it, select text, and copy from it.
Is my scanned document uploaded anywhere?
No. The whole recognition runs inside your browser tab using WebAssembly. Your file never leaves your device, which matters because scans are often contracts, IDs, or medical records.
Which languages are supported?
English, Spanish, French, German, Italian, Portuguese and Dutch. Choose the language before processing — it makes a real difference to accuracy. The list stops there because a PDF's searchable layer here stores Latin-1 characters only; offering, say, Polish or Turkish would recognise the text and then silently lose its accents on the way into the file, which is worse than not offering it.
Why is it taking a while?
Recognition happens on your own device rather than on a server, so speed depends on your machine — roughly 3 to 10 seconds per page. Larger scans and phones take longer. You can keep the tab open in the background.
What if some pages already have text?
Those pages are detected and left alone. Re-recognising a page that already has a good text layer would only stack a worse copy on top of it, so the tool skips them and keeps the original text.
Will the OCR be perfect?
Not always. Clean, straight scans at 300 DPI or better read very accurately. Faint photocopies, handwriting, heavy skew, and unusual fonts are where mistakes creep in, so check anything critical.