Image to Text (OCR)
Pull the text out of photos, screenshots and scanned PDFs — on your device.
Your files never leave your device
Drop files here
or click to choose files
First use downloads the OCR engine (about 7 MB) from this site — not from any third party — and it then runs entirely on your device. English recognition; more languages coming.
How it works
This tool reads the text out of images — photos of documents, screenshots, whiteboard shots — and scanned PDFs, and gives it back to you as plain text you can copy or download. Scanned PDFs are processed page by page, with each page's text labeled, so a multi-page scan becomes one clean text file.
Unlike virtually every online OCR service, nothing you drop here is uploaded. The recognition engine (Tesseract, compiled to WebAssembly) is served from this site and runs inside your browser; on first use it downloads about 7 MB of engine and language data — from us, not from a third-party CDN — and after that your images are processed entirely on your device.
Recognition currently supports English text and works best on clear, well-lit, straight-on images of printed text. Handwriting and low-resolution photos will produce rough results — that's a limitation of on-device OCR generally, and we'd rather be honest about it than upload your documents to something more powerful.
Frequently asked questions
Which languages are supported?
English for now. Additional languages (including Chinese) are planned — each adds several megabytes of language data, so we're adding them carefully rather than all at once.
Why is the first recognition slow?
On first use, your browser downloads the OCR engine and English language data (about 7 MB total) from this site, then keeps it cached. Subsequent recognitions start much faster.
Are my images or PDFs uploaded for recognition?
No. The engine itself is downloaded from this site to your browser, and all recognition happens on your device. Your images never leave it — verifiable in the network tab.