PDF Troubleshooting

OCR PDFs: Make Scanned Documents Searchable

Published 2026-08-23 · FreePDFToolbox

A scanned PDF is just a picture until OCR gives it a text layer. Here's what OCR can and can't do — and how to run it privately.

Scanned PDFs look like documents but behave like photos: you can't select, search or copy the text. OCR fixes that by recognizing the characters in the image.

What OCR actually does

Optical character recognition analyzes the shapes in each page image and maps them to letters and words. The result is a text layer on top of the scan.

  • Text becomes searchable — you can Ctrl+F the document.
  • Text becomes selectable — you can copy and paste it.
  • The visual page doesn't change — the scan stays as it is.

How accurate is OCR?

Scan qualityTypical accuracy
Crisp, straight, high contrast98%+
Slightly rotated or shaded90–97%
Noisy, low resolution70–90%
HandwritingUnreliable

Tips for better results: scan at 200–300 DPI, straighten pages, keep contrast high, avoid creases and shadows.

Run OCR privately

The OCR PDF tool runs the Tesseract engine in your browser via WebAssembly. The language model is bundled with the site — no cloud API, no uploads.

  1. Choose the scanned PDF.
  2. Select resolution (200 DPI balances speed and accuracy).
  3. Click RUN OCR and download the recognized text.

What you get

The tool outputs a TXT file of the recognized text. For a searchable PDF (text layer inside the PDF), OCR needs to write back into the document — a capability beyond this tool — but the extracted text covers quoting, analysis and search.

Frequently asked questions

Is OCR really private?

Yes — the recognition engine runs locally. Pages are never transmitted.

Why is OCR slow?

Recognition is CPU-heavy. Higher resolution means more pixels to analyze. Large documents take a minute or two.

Does OCR work with other languages?

The current build bundles the English model; other languages may be added later.

Run OCR now with OCR PDF.