OCR — Extract Text

Read Text From Any Scan or Photo — Without Uploading It

OCR (optical character recognition) turns a picture of text — a scanned contract, a photographed receipt, a screenshot, a page from a book — back into real, selectable, searchable text you can copy and edit. This tool does it with the open-source Tesseract engine compiled to WebAssembly, which means the recognition runs on your own device. The one-time language model downloads from a public CDN, but your document is never sent anywhere. That makes it safe for exactly the material people are most nervous about scanning into a random website: IDs, bank statements, medical letters, and legal paperwork.

How to Extract Text From a Scan or Image

1. Add your file. Drop in a photo, screenshot, or scanned PDF. Images (JPG, PNG, WebP) and multi-page PDFs are both supported.
2. Pick the document language. Choosing the right language dramatically improves accuracy. You can even combine two (for example English + French) for mixed documents.
3. Click Extract Text. The first run downloads the language model to your browser (a few seconds); after that it is cached. A progress bar shows recognition page by page.
4. Copy or download. The recognized text appears in the box below — copy it to the clipboard or download it as a .txt file. Your original file is never modified.

What OCR Is Good At (and Its Limits)

Accuracy is highest on clean, high-contrast documents: a straight 200–300 DPI scan of printed text will usually come back almost perfect. Accuracy drops on blurry phone photos, skewed pages, low light, unusual fonts, or handwriting (Tesseract reads printed text far better than cursive). For the best result, use the sharpest, most level, highest-contrast image you can, and pick the correct language. If a scan is faint, increasing contrast before running OCR helps a lot.

Common Uses

• Make a scanned PDF searchable: pull the text out of a scan so you can search, quote, or reuse it.
• Digitize receipts and invoices: get the numbers and vendor details into text you can paste into a spreadsheet.
• Capture text from a photo: a whiteboard, a book page, a sign, a slide — anything you photographed.
• Accessibility: convert an image-only document into text a screen reader can announce.

How It Works & Why It's Private

When you click Extract Text, the tool loads the Tesseract OCR engine (WebAssembly) and the language model into your browser. For a PDF, each page is rendered to a canvas locally with pdf.js; for an image it is read directly. Tesseract then analyzes the pixels on your device and returns the recognized characters. The only network traffic is downloading the engine and language files — the same for every user — never your document. You can confirm this on our Proof of Privacy page: load the tool, switch off your network after the model has cached, and OCR still works completely offline.