Scanned PDF to text (OCR)

Get the words out of a scanned letter, a photographed receipt or an old typed report. Pick a page, press Extract text, copy the result.

Your PDF stays on your device. Only the page you choose is sent, as an image, for recognition, and it isn't kept.

Loading the editor…

How to extract text from a scanned PDF

  1. Open the PDF above. The Scan to text panel opens beside it (below it on a phone).
  2. Choose the page number and press Extract text.
  3. Read and fix the result in the box; it is editable.
  4. Copy it, save it as .txt, or add it to that page's notes.

Is my PDF a scan? A quick test

Choose Edit existing text in the toolbar and tap a word. If it opens for editing, the page already has real text and you don't need OCR: edit it directly. If you get "No selectable text there", the page is an image and this is the right tool.

What comes back

  • Text in reading order, with line breaks and headings kept.
  • Lists as lists, and tables as simple pipe tables you can paste into a spreadsheet or Markdown editor.
  • [illegible] where the model can't read a word, instead of a guess.

Getting the best result from a phone scan

  • Shoot the page flat, in daylight, filling the frame; shadows across the text cause the most errors.
  • Use your phone's document scanner mode (Notes on iPhone, Google Drive on Android): it straightens and sharpens the page before saving as PDF.
  • For multi-column layouts such as newsletters, check that columns came back in the right order.

Questions people ask

What is PDF OCR?

OCR (optical character recognition) turns a picture of text, such as a scanned page or a phone photo saved as PDF, into real characters you can copy, search and edit. Here it is done by an AI vision model, which also copes with tables, columns and many handwriting styles.

Is my scanned PDF uploaded?

The PDF itself is not. When you press Extract text, only the one page you chose is rendered to an image in your browser and sent to the vision model through Vercel AI Gateway. Nothing is stored after the text comes back. Other pages, form answers and your notes are not sent.

Does it make my PDF searchable?

Not inside the PDF: it gives you the text to copy, save as .txt or keep as page notes. Embedding an invisible text layer into the PDF is what desktop OCR apps such as Adobe Acrobat Pro do.

How accurate is it?

Very good on clean printed scans; weaker on faint copies, tiny print, unusual fonts and messy handwriting. Unreadable words are marked [illegible] rather than guessed. Always check numbers, names and dates against the page.

Is there a page limit?

One page per request, and a fair-use limit of eight pages every ten minutes, because each page costs real money to process. It is free to use.

Can I then edit the scanned page?

Yes. Cover the old words with Whiteout and type the corrected text with the Text tool, copying from the extracted text.

Other things you can do here