Turn an image into editable text

Sometimes the text you need lives inside an image: a screenshot with no copy button, a photo of a whiteboard, a scan of a receipt, a slide someone shared as a PNG. This tool runs Tesseract.js — the browser build of the industry-standard Tesseract OCR engine — right in your browser to pull the text out.

Because it runs client-side, your image is never uploaded. That matters when the picture contains anything sensitive: a bank statement, a medical record, a login screen, a private conversation.

Supported languages

• English — best-in-class accuracy on printed text.
• Arabic — right-to-left with diacritic support.
• French / Spanish / German — accent-aware.
• Chinese Simplified — for typed and clear-print output.

Each language downloads a small trained-data file the first time you use it (2-20 MB). Your browser caches these, so switching languages later is instant.

Tips for the best results

• High-contrast, high-resolution images work best. A blurry phone photo of a book page gives worse results than a clean scan.
• Straighten and crop before uploading if you can — Tesseract does its own de-skewing but you get a boost from doing it first.
• For handwriting, expect much worse results than typed text. Tesseract is not designed for cursive handwriting.
• For long documents split across many pages, run one page at a time and paste the results together.