OCR PDF

Add an invisible text layer to a scanned PDF so that it can be searched and copied. Page images are not changed, and pages already holding more than a little text are skipped unless you ask. English only for now.

Good to know

OCR can misread, especially on poor scans, handwriting and small print. Check anything that matters against the page image.
The engine and English language data load from this site when you run it, and the document stays in your tab.
Characters the embedded font cannot show are replaced in the text layer, and you can cancel at any page.

Your files stay on your machine

This tool runs entirely inside your browser tab. Nothing is uploaded, stored, or sent to a server — close the tab and every trace is gone.

These are the free ones — see the full suite

How big a file can it take?

There is no upload, so there is no upload limit — but the work happens in this tab, which means your browser’s memory is the budget. Nothing here will refuse a file for being large; a big enough one simply gets slow and, eventually, reloads the tab. The figures below are rules of thumb on a desktop browser, not enforced caps. On a phone or tablet, expect a good deal less.

OCR is slow: expect several seconds a page on a desktop, and much longer on a phone. Up to about 200 pages in one sitting is reasonable.
Each page is rendered at high resolution while it is read, so memory use follows page size. Large-format sheets are the heaviest.

Password-protected PDFs are declined whatever their size — unlock the file first. If a run does die on a long document, splitting it into volumes and working through them one at a time is the reliable way round it.