How it works

This runs entirely in your browser using Tesseract.js — a WebAssembly build of Google's Tesseract OCR engine. Your image is never uploaded. Most sites make you send scans to their server for this; you don't have to.
The honest tradeoff: the language model downloads once (a few MB) on first use, and recognition is slower than a server would be — expect several seconds per page, longer on an old phone. That's the price of not uploading your documents.
Accuracy depends heavily on input quality. Clean, straight, high-contrast scans read very well. Photos taken at an angle, blurry images, unusual fonts, and handwriting read poorly. The confidence score tells you how much to trust the output — always proofread it.
13 languages are supported. Pick the right one before running: OCR with the wrong language model produces nonsense.
For multi-page PDFs, use PDF Split first, then run each page through here.