OCR PDF: make a scanned PDF searchable

Before you start: what this tool can and cannot do

English only. This tool recognizes English text. No other OCR languages are available at launch, so text in other languages will be missed or come out wrong.
OCR can make mistakes. Recognition is not guaranteed to be accurate. Small or blurry print, handwriting, unusual fonts, low-quality scans, tables, and complex layouts can produce wrong or missing words. Check important text before you rely on it.
Large documents can reach your browser's memory limit. Each page is turned into a large image while it is read, and the whole PDF stays in this tab's memory. Very large OCR documents (hundreds of pages or very large files) can make the browser slow down or stop the tab, especially on phones. If that happens, close other tabs, or split the PDF and run OCR on smaller parts.
Your pages are not re-drawn. The original scan stays exactly as it was. Pages that already have selectable text are kept as they are, and the page count and order do not change.
OCR takes a few seconds per page. The first time you start, the OCR engine and English model (about 7 MB) are loaded from this website. Your PDF and its text are never sent anywhere.

PDF files only. Your file stays on this device.

Recognition language: English (the only language available)

Choose a scanned or image-only PDF from your device, or drop it on the box above.
Press Start OCR. Progress is shown page by page, and you can cancel at any time. Nothing happens until you press the button.
Download the searchable PDF. It looks like the original, but you can now search for words, select them, and copy them in a PDF viewer.

Everything happens in this browser tab. The PDF, the page images made while reading it, the recognized text, and the new PDF are not sent anywhere.