Skip to main content
OS Authority

Make a scanned PDF searchable with OCR

Recognize printed English on scanned pages, review each line, and add invisible searchable text while keeping the original PDF pages and their appearance.

Your PDF and recognized text stay in browser processing workers on this device. Their contents are never uploaded or included in analytics.

How it works

  1. Choose an ordinary, unencrypted PDF containing printed English scans.
  2. Enter the pages to recognize, or leave the selection empty for all pages within the device limit. Pages with existing text are skipped.
  3. Run recognition, compare each page preview with its transcript, and correct words without adding or removing lines.
  4. Create the searchable PDF, download it, then test search and copied text in your PDF reader.

Scanned pages versus digital text

A digital PDF often already contains selectable text. A scan may look the same but store only an image. This tool checks selected pages and skips any page with detected text to avoid adding a duplicate layer. In a 2-page document with a digital cover and scanned letter, the cover stays unchanged and the scanned letter can gain searchable text. Existing incomplete or incorrect text layers are not replaced.

Review the hidden text layer

Searchable text is invisible: correcting a recognized word changes search and copying, not the printed pixels of the scan. Keep the original line count during editing so each line stays associated with its recognized position. Selection boxes and reading order are estimates, especially with columns or unusual layouts. Check names, amounts and dates against the source; a confidence score of 90 does not establish 90% accuracy.

Understand the output and its limits

The PDF keeps its original visual page streams, CropBox and page rotations. Only reviewed recognized pages receive added text. Device limits bound source pages, raster pixels and output bytes; split a long document first if necessary. Forms, annotations, advanced interactive structures and encrypted PDFs are unsupported. Modifying a signed document can invalidate signatures; this tool is not intended for signed files. It does not produce certified PDF/A or a tagged accessible PDF.

Frequently asked questions

Can I search and copy the result?

The added invisible text can be searched and copied in a reader that supports searchable PDFs. OCR mistakes and reading-order issues can remain, so test a known word and inspect copied text after downloading.

What happens to pages that already contain text?

They are always skipped and preserved. The tool does not replace an existing text layer, even if that layer is incomplete or inaccurate. This policy avoids duplicating text in the output.

Does the output qualify as PDF/A or an accessible PDF?

No compliance conversion or certification is provided. A searchable layer alone does not establish PDF/A, tagging, accessible reading order or suitability for a court, archive or filing system.

Can it recognize handwriting or non-English scans?

This version supports printed English. Handwriting, other language models, table reconstruction and guaranteed multi-column reading order are unsupported.

Can I change the visible scan while reviewing text?

No. Review edits affect the invisible searchable layer only. Keep one edited line for each original recognized line; use another PDF editor if you need to change visible page content.

Last reviewed: October 11, 2026

Related tools