NimboOCR PDF reads each scanned page with Tesseract, the open-source text recognition engine, running in your browser. It adds an invisible text layer on top of the original pages: the document looks exactly the same, but you can now search it, select words and copy them. You also get the recognised text as a .txt file.
How it works
- Drop the scanned PDF.
- Choose the language of the document, and English too if the text mixes both.
- Select Make searchable, then download the searchable PDF or the text.
Supported files
Input: one PDF with scanned or photographed pages. Output: the same PDF with a searchable text layer, plus a .txt file.
Up to 400 MB per PDF on a computer, 150 MB on iPhone and iPad. Larger files depend on your device's memory.
When to use it
- Search inside scanned contracts, invoices or course notes.
- Copy a paragraph from a scan instead of retyping it.
- Make archived paper documents findable by your computer's search.
Quality and limitations
- Recognition is never perfect: check names, numbers and amounts.
- Handwriting is recognised poorly; printed text works best.
- Long documents take time, especially on a phone: count a few seconds per page.
Common issues
- “No text was found on these pages.”
- Check the language you chose, and that the scan is sharp and upright. Photos taken at an angle are recognised poorly.
- The first run is slow.
- The language model (2 to 4 MB) is downloaded from this site the first time, then reused by your browser.
Private by design
This tool runs entirely in your browser with JavaScript. Your file is read and processed on your device and is never uploaded to a server. Closing the tab discards everything.
Frequently asked questions
Which languages are supported?
The 19 languages of this site, from English, French and German to Japanese, Korean, Traditional Chinese and Arabic. You can add English to any of them for mixed documents.
Is my scan uploaded?
No. Text recognition runs in your browser, on your device.
Does it change how the pages look?
No. The original pages stay as they are; an invisible text layer is added on top.