Image to Text (OCR)
Pull the text out of a photo, a screenshot or a scan, then fix it right here.
About this tool
A receipt you photographed, a screenshot of a message, a page from a book, a road sign, an old scanned letter: the text is right there, yet impossible to copy. Optical character recognition brings it back to life. Drop the image and the engine reads the page, finds the lines and hands you the text in an editable box, with the number of words it recognised and an honest confidence score. Read it over, correct whatever needs correcting, then copy it in one click or download a clean .txt file.
Everything happens inside your browser: the recognition engine (about 3 MB) and the language model are downloaded once and then cached, while your image never reaches a server. That is what matters for a payslip, a bank statement or a contract you would rather not upload to a stranger's website. The recognition language is preselected from the site language and can be swapped for one of twenty others, and badly lit photos of documents are automatically put through a second pass that cleans up the contrast.
Frequently asked questions
Are my documents uploaded anywhere?
No. Only the recognition engine and the language model travel to your browser, once. The image itself is read on your own machine and never leaves it: no account, no storage, nothing to delete afterwards. You can even go offline after the first run.
Which languages can it read?
The site language is preselected (English, French, Spanish, German, Portuguese, Italian) and the picker offers around twenty more: Dutch, Polish, Romanian, Czech, Swedish, Turkish, Greek, Russian, Ukrainian, Arabic, Japanese, Simplified Chinese, Korean, Vietnamese. There is also a combined option for mixed documents, and each model is downloaded only once.
Does it read handwriting?
No, and it is only fair to say so plainly: this engine is trained on printed text. Handwritten notes, signatures or very decorative fonts will come out unusable. Invoices, contracts, books, receipts, screenshots and scanned paperwork, on the other hand, are exactly what it is built for.
How do I get the best result from a photo?
Shoot the page flat and roughly straight, with even light and no shadow falling across the text, and aim for a sharp picture at least 1000 pixels wide. The tool already cleans up poorly lit photos on its own, but a blurred or heavily tilted image stays unreadable for any OCR engine. The confidence score tells you when to proofread closely.
Is it really free, with no sign-up?
Yes — the tool is free, needs no account and has no usage limit. Processing happens locally in your browser, so your data never leaves your device.