StaplePDF

OCR PDF: make a scan searchable

A scanned PDF is only pictures of pages. Text recognition (OCR) reads the words in them, so you can search the PDF, and select and copy its text. It all happens on your device.

or drop it here

🔒 Your files stay on your device. Nothing is uploaded.

How to OCR a PDF

  1. Choose your PDF. It opens in the StaplePDF workspace at Recognize text (OCR).
  2. Pick the language the document is written in, then select Recognize text. Each page takes a few seconds. The first time, your browser downloads about 4 MB to do it; after that it's ready straight away.
  3. Search with Ctrl+F (⌘F on a Mac), or select and copy text, to check the result. Then save the searchable PDF.

What changes in your PDF

The pages look exactly as they did. The recognized words are added as invisible text, placed over the words in the scan, which is how searchable PDFs from scanners and Adobe Acrobat work too. Any PDF reader can then search it, and you can select and copy the text. Pages that already have text are left alone, unless you choose otherwise.

Private by design

Most online OCR services upload your document to read it on their servers. StaplePDF runs the open-source Tesseract engine in your browser instead: the first time, your browser downloads the engine and the language data (about 4 MB for English, less for the others) from staplepdf.com, and from then on your document is read on your device and never sent anywhere.

Questions

How accurate is it?

Clear scans of printed text at 200 to 300 dpi are usually read almost perfectly. Handwriting, very small or faint print, and photos taken at an angle are harder. Words it's unsure of are left out rather than guessed.

Which languages does it read?

English, Spanish, French, German, Portuguese and Italian.

My scan is sideways. Will it work?

Turn the pages the right way up first, with the rotate buttons in the workspace (or Rotate PDF), then recognize the text.

Can I edit the text afterwards?

OCR makes text searchable and copyable; the page itself is still the scanned image. To change wording, cover it and add new text with Edit in the workspace.