OCR PDF
Make a scanned PDF searchable and copyable. The pages look exactly the same.
When to use it
Use it for scanned contracts, letters or books, so you can search the PDF, select text and copy it.
How it works
- Choose your files, or drop them onto the box above. You can add several at once.
- Choose the document language, then run. Every page is read on your device and an invisible text layer is added on top of it.
- Press the Make searchable button.
- Download the result, or add a next step without downloading.
Supported formats
- Input
- PDF, PNG, JPG, WebP, BMP
- Output
- Files
- One or several; each file is processed on its own
Several results can be downloaded one by one or together as a ZIP.
Where your file goes
This tool runs in your browser: the file is processed on your device and is not uploaded. The Network tab of your browser's developer tools shows this.
On first use, your browser downloads the code this tool needs from this site. Your file is not sent anywhere.
What is kept, and what is not
The original pages are kept untouched — only invisible text is added. Recognition is never perfect: words the engine is unsure of are counted so you know what to check. Urdu Nastaliq is in beta.
Limits
- Recognition works best on clear, straight scans. Handwriting is not supported.
- Files up to 512 MB each.
- PDF documents up to 5,000 pages.
- Images up to 100 megapixels.
- The work happens in your device's memory, so a phone may fail on a file that a computer handles.
If something goes wrong
Many words are recognised wrongly
Check that the document language is set correctly and try the accurate quality setting. Low-resolution or tilted scans reduce accuracy.
The PDF asks for a password
Remove the password first with Unlock PDF (you need to know it), then try again.
The page becomes slow with a very large file
Close other tabs, or use a computer instead of a phone.
My file is not accepted
Check the formats listed above. A damaged file, or one with the wrong extension, is refused with a message saying why.
Questions
Which languages can it read?
The languages listed in the language option: English, Arabic, Urdu, Bengali, Azerbaijani and Russian, and English with Arabic together.
Does the scan look different afterwards?
No. The pages stay exactly as they were; an invisible text layer is added on top.
Continue with the result
The result can go straight into another tool on this page, without downloading it and uploading it again: