OCR2textRuns on your deviceAbout
OCR2text exists because the alternatives are either an ad-covered upload box that emails your invoice to a server in another country, or a subscription app you have to install. Neither is a good answer to “what did this page say?”
Browsers can now run a real OCR engine at a genuinely useful speed. Once that is true, uploading a private document to have it read back to you stops making sense. Everything else follows: no account to create, no storage to trust, no per-page cost, and no reason not to work offline.
It is not a handwriting recogniser — the engine is built for printed text, and neat handwriting is the most you should hope for. It is not an accounting tool: receipts get read, not categorised. And it is not an AI writing assistant. Nothing here rewrites your words, corrects your spelling or paraphrases your document.
Nothing, and there is no crippled free tier waiting to be upgraded. Because recognition runs on your device there is no GPU server, no OCR API bill and no document database to pay for — so basic OCR can stay free without anyone having to be clever about it.
OCR2text is made by Techvora. Recognition is powered by the Tesseract project, and PDF handling by Mozilla’s PDF.js — both open source, both running on your machine rather than ours.
The workspace is keyboard navigable, the text editor is a real editable surface with native undo, redo and spellcheck, focus is always visible, and every animation stops when your system asks for reduced motion. If something here does not work for you, that is a bug worth telling us about.
Small, focused tools that do one job properly. Have a look at techvora.net, or start with the workspace.
No account, no upload, no waiting for a server. Open the workspace and drop a page in.