How to Extract Text from a Scanned PDF (Free, No Upload)

Some PDFs let you select and copy text; others fight you. The difference is what’s inside: a digital PDF contains actual text data, while a scanned PDF contains photographs of pages — to a computer, each page is one big image.

First: which kind do you have?

Open the PDF and try to select a sentence with your mouse.

• Text highlights line by line → digital PDF. Extraction is trivial and instant.
• You get a rectangular “marching ants” selection, or nothing → scanned PDF. You need OCR.
• Some pages select, some don’t → mixed document, common with contracts where a signed page was scanned back in.
• You can select, but the copy is garbage (□□□, punctuation soup, or letters in the wrong places) → the text layer is broken. Treat it like a scan.

You don’t have to check by hand. The PDF to text tool inspects every page and picks a text-layer read or OCR automatically. If the automatic pass still looks encoded-wrong, use Re-extract with OCR to force every page through the visual path.

Extracting text from a scanned PDF

1. Open the PDF to text tool.
2. Drop your PDF onto the page.
3. Watch the per-page progress. Pages with a usable text layer are read directly (fast); scanned or broken pages are rendered and run through OCR (a few seconds each).
4. Copy or download the combined text.

Need one paragraph, not the whole file? On the preview, drag a box or click a block, then extract that region. The rest of the document stays in the full-result view.

Everything happens in your browser: pdf.js parses the file, and scanned pages are OCRed by a neural engine running as WebAssembly. The file is never uploaded — which matters, because scanned PDFs are so often contracts, invoices, and records.

OCR on those pages reads printed English, German, French, Japanese and Korean, plus Chinese. Digital text-layer pages are not limited by that list: whatever glyphs the PDF actually stores can be copied out.

What the tool cleans up for you

Raw PDF extraction has a few classic annoyances:

• Headers, footers and page numbers repeat on every page. Lines that recur across most pages are stripped automatically.
• Hard line breaks — PDFs store text as positioned fragments, so paragraphs arrive chopped into layout lines. Toggle Clean up text to join those lines and remove end-of-line hyphenation.
• Floating punctuation — Chinese and Japanese quotes, parentheses and 【】 often sit on a different baseline than the letters. A naive sort parks “ ” or ( ) at the start of the line. The extractor puts those marks back around the phrase.
• Tables — column-aligned text on a digital page is rebuilt as a table; scanned pages that look grid-like go through a local structure model.

Getting better results from rough scans

OCR quality follows scan quality. If you control the scanning step:

• Scan at 300 DPI or higher — 150 DPI receipts are where OCR goes to die.
• Keep pages straight; heavy skew hurts recognition.
• Prefer black-and-white or grayscale over color for text documents; it boosts contrast.
• If a scan is very poor, re-photographing the paper with your phone in good light sometimes beats the original scan.

More on capture quality in the OCR results guide.

A note on file size

Because processing is local, big documents cost your device time, not a server’s. Digital pages are read at hundreds of pages per minute; scanned pages take a few seconds each. A 300-page digital report is quick; a 300-page pure scan will take a while — leave the tab open and let it run.