OCR PDF
Read the words out of a scanned PDF — the kind that is really just photographs of pages, with no text a computer can select.
- Runs in your browser No upload — your files never leave your device.
- No sign-up, no watermark Free to use, no account needed.
- Files never uploaded Close the tab and your files are gone from memory.
- 5,400+ tools, one site File, unit and currency tools in one place.
Read the words out of a scanned PDF — the kind that is really just photographs of pages, with no text a computer can select. This tool renders each page as a picture and reads it with a built-in OCR engine, running entirely in your browser with its English language data stored on this site, so nothing is uploaded anywhere. You get a JSON file with the text of every page, each page's confidence score, and the whole document's text joined together. English is the default language, and other languages can be chosen with their language codes, such as eng+fra for a document in English and French. By default only the first ten pages are read, because OCR takes seconds per page and a long document would take a long time; you can raise the limit to fifty pages. Proofread the result: clean printed text reads almost perfectly, while handwriting, skewed scans and unusual fonts can misread.
How to use OCR PDF
- Choose your file drop it onto the box above or browse your device.
- Set the options adjust any settings below the box, if the tool has any.
- Click Convert the conversion runs entirely in your browser — nothing is uploaded.
- Download the result save the converted file to your device.
- Convert another clear the tool and choose a new file — no sign-up, no limits.
Format facts
| Input | A file from your device. |
|---|---|
| Engine | Built-in browser engine |
| Category | PDF Tools |
Frequently asked questions
What kind of PDF does OCR work on?
Scanned PDFs and PDFs made of page photographs — files where the text is a picture, not selectable text. If your PDF already has selectable text, you do not need OCR at all: copy the text directly instead of reading a picture of it.
Which languages can it read?
English (eng) by default. Other languages use OCR language codes, and two can be combined with a plus sign, for example eng+fra for English and French. An invalid code is refused with a plain message.
How many pages will it read?
The first 10 pages by default. OCR takes seconds per page, so reading a very long document would take a long time. You can raise the page limit up to 50 pages; if the file is longer, the result says so and only the first pages are included.
How accurate is the result?
Very good on clean, straight, printed text — the test documents read back exactly. Handwriting, skewed or blurry scans, tiny print and unusual fonts can misread, so always proofread the result before relying on it. Each page carries a confidence score as a rough guide.
What do I download?
A JSON file with the full text, plus each page's text and confidence score separately. Open it in any text editor, or use the text shown on the page after converting.
Was this tool helpful?
Thanks — your feedback helps us improve this tool.
Sorry it didn’t help — tell us what went wrong and we’ll fix it.