Add a photo, screenshot or scanned PDF. You can paste a screenshot straight from your clipboard with Cmd/Ctrl+V. The file is opened by your browser, in this tab — there is no upload step, anywhere.
PDF pages are drawn to an offscreen canvas by pdf.js — the same engine Firefox uses to display PDFs. Images are decoded and, if huge, scaled to an ideal size for recognition.
Each bitmap is handed to Tesseract — a 25-year-old open-source OCR engine compiled to WebAssembly — running in a Web Worker so the page never freezes. Its LSTM neural network recognises the text line by line.
The recognised text appears below, per page and combined. Copy it,
download it as .txt, or share it. It never leaves your
machine.
Textlift can also rebuild your document as a searchable PDF — the original pixels with an invisible text layer underneath, so you can select, copy and Cmd/Ctrl+F it in any PDF viewer.
Textlift pulls the text out of images, screenshots and scanned PDFs without uploading them anywhere. Most "image to text" sites send your file to a server — and OCR inputs are exactly the documents you least want on someone else's server: IDs, invoices, contracts, prescriptions. Textlift runs the OCR engine itself, compiled to WebAssembly, inside your browser.
It reads printed text in 12+ languages. Handwriting is not supported — that's a limit of the Tesseract engine, and it's better to say so than to pretend.
Built by Ben Richardson. Explore the full catalog of privacy-first tools at hub.benrichardson.dev.
Source code: github.com/ben-gy/textlift (MIT).