Extract text from scanned PDFs using Tesseract OCR. Upload a scanned PDF, choose language, and get the recognized text for each page. Runs entirely in your browser — no server.
Drop a scanned PDF on the left, pick a language, then Run OCR — the recognized text for each page appears here with per-page copy and a full .txt export.
OCR (optical character recognition) turns the pictures of words inside a scanned PDF back into real, selectable text. This tool renders each page of your PDF with pdf.js and runs Tesseract.js — a fully open-source OCR engine — directly in your browser to recognise the characters, so you get copy-pasteable text from documents that were previously just images.
It supports nine recognition languages, including English, French, German, Spanish, Italian, Portuguese, Simplified Chinese, Japanese and Arabic. People use it to:
Everything runs locally in the browser — your PDF is read into memory and processed on your own device. Nothing is uploaded to a server, so even sensitive scans stay private on your machine.
OCR (optical character recognition) reads the images of letters inside a scanned page and converts them into real text. This tool renders each PDF page to a canvas and runs the Tesseract.js OCR engine on it, giving you selectable, copyable text from a document that was just a picture.
A scanned (image-based) PDF is a photo or scan of a page, so its words are pixels with no underlying text. A digital PDF was created from a document and already contains real text. OCR is meant for scanned PDFs — for digital ones you can extract the text directly without recognition.
You can choose from nine languages: English, French, German, Spanish, Italian, Portuguese, Simplified Chinese, Japanese and Arabic. Pick the one that matches your document before running OCR for the most accurate results.
Tesseract is very good on clean, high-contrast, straight scans of printed text. Accuracy drops on blurry, skewed or low-resolution scans, decorative fonts and handwriting, where a cloud OCR service like Google Vision or AWS Textract may do better.
OCR is processed one page at a time on your own device, so large or many-page PDFs take longer. The first run also downloads the chosen language model (roughly 5–20 MB); after that the model is cached and subsequent runs are faster.
Yes. Each page has its own Copy button to put that page's text on your clipboard, and the Download Full Text (.txt) button saves the combined text of every page as a single plain-text file.
No. The entire process — rendering pages and running Tesseract — happens locally in your browser. Your PDF is never sent to a server, so even confidential documents stay private on your device.