Extract Text from Scanned PDFs & Images (OCR)
Turn a scanned PDF, a photographed page or a screenshot into selectable, copyable text. Upload the file below, pick the document's language, and download the recognized text as a .txt file.
Runs on your device — your file is never uploaded (up to 50 MB).
Unlike almost every OCR site, this one never sees your document: the recognition runs entirely inside your browser on your own device. Nothing is uploaded — not to us, not to anyone.
How it works
The tool uses the Tesseract OCR engine compiled to WebAssembly, running in a background thread of your browser. Scanned PDFs are first rendered page by page at high resolution, then each page image is recognized character by character using a language model for the language you pick. The first run downloads that language model (a few megabytes of static data); your file itself is never transmitted.
Because everything runs on your device, speed depends on your hardware — a phone takes noticeably longer than a laptop — and accuracy depends on the scan: sharp, straight, well-lit pages read nearly perfectly, while skewed or low-resolution scans produce more errors.
How to extract text
- Upload a scanned PDF, JPG, PNG or WEBP above.
- Select the language the document is written in — this matters for accuracy.
- Click Extract text and watch the per-page progress.
- Download the .txt file, or open it and copy what you need.
Limitations to know about
- OCR output is plain text: layout, columns, tables and images are not reconstructed.
- Accuracy depends on scan quality and the right language choice; handwriting is generally not readable.
- PDFs are limited to 20 pages per run — split larger documents first with the Split PDF tool.
- Recognition runs on your device, so very large documents take time on slower hardware.
Privacy and security
Your file is uploaded over HTTPS, converted entirely in memory and streamed straight back to your browser. It is never written to disk, never logged and never retained after the conversion — there is nothing to delete because nothing is stored. Files are validated by extension, type and content signature, and uploads are never executed. Read more in our privacy policy.
Frequently asked questions
Is my document uploaded to a server for OCR?
No. The OCR engine runs inside your browser via WebAssembly, so the document never leaves your device. The only network activity is a one-time download of the language model itself — static data that reveals nothing about your file.
Which languages can it read?
English, Spanish, French, German, Portuguese, Italian, Arabic, Simplified Chinese, Indonesian and Turkish. Pick the language of the document, not your interface language — recognition uses language-specific models.
Can I turn a scanned PDF into an editable Word file?
In two steps: run OCR here to get the text, then paste it into Word or convert the .txt onward. The PDF to Word converter handles text-based PDFs directly, but scans need this OCR step first because a scan contains pictures of letters, not text.
Why is the recognized text wrong in places?
OCR accuracy tracks scan quality. Blurry photos, small fonts, skewed pages and the wrong language setting are the usual causes. Rescanning at higher resolution, cropping to the page, and double-checking the language selection fix most of it.