Files never leave this device

Text recognition

Turn scanned pages into text you can search.

Recognize the text of scanned pages directly in your browser — English or Spanish — and download it as plain text or as a searchable copy of the PDF. Nothing is uploaded.

Reading files…

What it solves

Text you can search and copy, out of pages that were only pictures.

This tool recognizes text in scanned pages with Tesseract models served from this site and run in your browser — English and Spanish are available today. Choose the language and the pages, follow live per-page progress with the option to cancel, then download the plain text as a TXT file or a searchable PDF that layers invisible text over the original pages.

Recognition quality depends on scan quality: clean, straight, high-resolution scans read well, while faint or skewed ones lower the per-page confidence. Runs are limited to 20 pages at a time so this device's memory is not exhausted.

Common uses

Verify a contract revision

Compare the draft you sent with the version that came back and see exactly which pages were touched before signing.

Make an archive of scans usable

Recognize the text of English or Spanish scans and export searchable copies you can quote and search.

Pre-check a publication

Run the four accessibility checks on a report before sending it to specialist review, and fix the obvious blockers first.

Quick guide

From a closed document to clear answers.

01Open the PDF you want to understand; for a comparison, add both versions.
02Choose what to run: the text comparison, the recognition language and pages, or the four accessibility checks.
03Run the analysis and follow the progress page by page, on the pages themselves.
04Read the verdict on screen and, if you need it, download the text or the report as a file.

Limits worth knowing

Honest about what analysis can see.

01
The comparison reads extractable text: scanned pages without a text layer cannot be compared, and purely visual changes such as images or layout are not detected. Recognize scans with OCR first.
02
Text recognition currently offers English and Spanish, up to 20 pages per run — a deliberate limit so the device's memory is not exhausted — and its quality depends on the quality of the scan.
03
The accessibility result is a practical pre-check, not a PDF/UA or WCAG certification; a compliant publication still needs specialist validation.
04
Everything runs on this device, so very large documents are limited by the browser's memory, and every result is a new download that never overwrites the original.

Frequently asked questions

OCR for PDFs — FAQ.

01Does this tool upload my PDF?

No. The analysis runs in the browser and the recognition models are served from this site, so the document bytes never leave the device or reach any server.

02Why does it find nothing on scanned pages?

A scan is a photograph of a page: there is no text layer to read. Run OCR first to add one, then compare or check the recognized copy.

03How reliable is the recognized text?

It depends on the scan: clean, straight, high-resolution pages read well, while faint or skewed ones lower the per-page confidence the tool shows. Review the result before relying on it.

04Is the accessibility result an official certification?

No. It checks four practical prerequisites — tags, language, title and scan-only pages — and full PDF/UA or WCAG conformance requires specialist validation and manual review.

Related tools

Keep working on the document.

All PDF tools

01 / PDF

All PDF tools

Share this tool

Referencing this tool on a website?

Copy an attribution link. It is optional and does not change free access.

Useful next stepCompare PDF FilesCompare two PDFs locally by page and extracted text, identify changed pages and quantify added or removed words.