OCR PDF

How it works

A scanned PDF is usually a photo of each page: you can see the text, but you cannot search or copy it. Choose the PDF, language of the text, and quality, and the tool recognizes the letters on each page (OCR) and adds them to the file as an invisible text layer. The page looks exactly the same; the difference is that Ctrl+F, text selection, and copying now work, and screen readers can read the document.

The result depends on scan quality. Clean, straight pages with normal-sized text usually achieve near-perfect accuracy; small text, smudges, skewed photos, or poor lighting will cause errors. Do not trust recognized text blindly: if it is important (numbers, names, dates), verify it against the image. If a page turns out poorly, try higher quality.

What it does not do: it does not read handwriting; does not recognize table structure (cell text is read as loose lines); on multi-column pages the reading order may be mixed; does not detect rotated or upside-down pages (rotate them first with Rotate PDF); and only has models for Spanish and English.

The text layer is invisible and does not change the appearance of the page. Your original PDF is not modified: you get a copy with the same content plus the added text, which adds a few KB per page. Pages that already have text are skipped unless you check the option to read them too.

Everything happens on your device and nothing is uploaded. The first time, the page loads the recognition engine and language models from this same site (several MB). Recognizing a page takes from a few seconds to nearly half a minute depending on its size and the chosen quality, and on a phone it can take considerably longer. You can cancel at any time and your file remains untouched.

Order matters if you plan to sign: run OCR first, then sign with Sign PDF. Adding text to an already signed PDF causes readers to report that it was modified after signing, so this tool refuses signed PDFs. Password-protected PDFs cannot be processed either. If you also want to compress it, the order is compress, OCR, and sign last.

What is not preserved, and limits

• Accuracy depends on scan quality: small text, smudges, skewed photos, or poor lighting will cause errors. Verify important details against the image.
• Only models for Spanish and English are available. Does not read handwriting, does not recognize table structure, and on multi-column pages the reading order may be mixed.
• Slower on mobile devices, and very large PDFs (over roughly 100-150 MB, or with thousands of pages) may fail.
• Signed PDFs and password-protected PDFs are rejected for now.