Scan to text
Pull the text out of a scan, a photo of a page or a PDF that is really an image — with the recognition engine running in your browser rather than your document going to somebody's server.
What it does
Scan to text changes which pages a PDF contains, or how they sit, and leaves everything else alone. The page content is copied rather than re-encoded, so nothing is recompressed and no quality is lost between the file you drop in and the one you download.
When you need it
Typically when a document carries pages that should not be shared — an internal cover sheet, a duplicated scan, an appendix meant for someone else — or when a scanner produced pages sideways or out of order.
Limits and caveats
Your original file is never modified: you always download a new copy. Encrypted PDFs have to be unlocked before they can be edited, and a scan is still a scan afterwards — the pages move, the text inside them stays an image. Everything happens in your browser, so the file never leaves your device.
How to scan to text
- Pick the language the document is written in — it decides which model is downloaded.
- Drop the scan, photo or PDF.
- The engine downloads once, then reads the page. Check the text and correct anything it misread.
Frequently asked questions
- Why does it download a few megabytes the first time?
- Because the recognition model comes to your document instead of your document going to a server. It is about 3 MB for one language, downloaded once and cached by your browser after that. Every free OCR that starts instantly does so by uploading your file.
- The text has mistakes in it.
- It usually will, somewhere. Recognition is a guess, and the guess gets much worse on low-resolution scans, photographs taken at an angle, handwriting and unusual fonts. The result is editable for exactly that reason, and the confidence figure tells you how sure the engine was. Below 70% treat it as a starting point, not an answer.
- Can it read handwriting?
- Not reliably. The model is trained on printed text. Neat block capitals sometimes come through; ordinary handwriting does not.
- What about a PDF with several pages?
- Only the first page is read. A scanned PDF is a picture of a page, and reading a hundred of them in a tab would take longer than anyone will wait — so this does one page well rather than a whole document badly.
- Do my files get uploaded?
- No. This tool runs entirely in your browser, so the file never leaves your device.