Extract text from scanned / image-based PDFs.
OCR runs in your browser using Tesseract.js. Nothing is uploaded.
or drag and drop a PDF here
Select or drop an image-only or scanned PDF. If it has an open password, enter it on the page. For ordinary PDFs where you can already select text, Extract Text is faster and more accurate.
Click Run OCR. The first time you use the tool, the recognition engine (a few megabytes) is downloaded into your browser and then kept for later visits. Each page is then read locally. This can take a while on long or high-resolution files.
Recognised text appears in the box on the page. Read through it, then copy what you need. Recognition is never perfect, especially on faint, skewed or handwritten scans.
Optical character recognition runs in your browser with Tesseract.js. After the engine has been fetched, page images and recognised text stay on your device. Your scan is not uploaded to an online OCR service, which is the usual reason people look for a private tool like this.
The tool is set up for English and works best on clear English print. It is not a multi-language OCR pack in this version.
The recognition engine is downloaded into your browser on the first run (a few megabytes). After that, recognition itself is still the slow step on long files.
No. Once the engine is on your device, pages are read locally. We do not receive the file or the text.
Not reliably. It is meant for printed, scanned pages. Handwriting, stamps and faint copies will produce errors.
Use a straighter, higher-contrast scan if you can. Crooked photos, coloured paper and very small type all reduce accuracy. Always proof the text.
No. Use Extract Text, which copies the existing text layer and is faster and more faithful.