PDF Tools · Free · No signup

OCR a Scanned PDF — On Your Own Device

Turn scanned pages into real, copyable text without uploading a thing. This tool renders each PDF page locally, then runs the Tesseract OCR engine (compiled to WebAssembly) right in your browser to recognize the text — English, French, Spanish, German, Portuguese and Italian supported. You get a clean .txt file with one section per page, and your documents never leave your device.

Drop your scanned PDF here or click to browse

You can also paste from the clipboard (Ctrl+V)

Language The OCR model (≈2 MB per language) is fetched from our server once, then cached — recognition runs on your device.

🔒 100% private: your files are processed by your own browser and never leave your device.

How it works

Drop your scanned PDF into the box above.
Pick the document's language (it changes which OCR model is used).
Press Run OCR. Pages are rendered and recognized one by one, with live progress.
Download the .txt with all recognized text.

Features

100% on-device OCR

The Tesseract engine runs as WebAssembly in your browser — scans are rendered, recognized and saved locally.

Six languages

English, French, Spanish, German, Portuguese and Italian models, selectable per job.

Live per-page progress

See which page is being processed and how confident the recognition is, in real time.

Clean text output

One section per page in a UTF-8 .txt — ready to search, quote or paste.

No account, no upload

The engine and your pages stay on your machine. Nothing is sent to any server.

Supported formats. Accepts scanned or image-only .pdf files (text-based PDFs should use PDF to Text instead — it is instant). Outputs .txt (UTF-8).

Common use cases

Digitizing signed or stamped paper documents that were only scanned.
Making old invoices, receipts and letters searchable.
Extracting text from faxes and photocopied archives.
Reading printed articles when the source PDF has no text layer.

Limitations (honest ones)

The first use of a language downloads its OCR model (2–5 MB, hosted on this website) — after that it is cached by your browser. OCR is CPU work: expect a few seconds per page, so keep very long documents to batches of ~30 pages. Accuracy depends on scan quality: 200–300 DPI, straight and clean scans do very well; blurry photos, handwriting and exotic fonts do not. Output is plain text — layout, columns and images are not preserved.

Frequently asked questions

Is my document uploaded anywhere?

No. Pages are rendered and recognized entirely in your browser. Only the OCR language model itself is downloaded from this website the first time you use a language.

Why is OCR slower than PDF to Text?

PDF to Text reads embedded text instantly. OCR has to actually look at pixels and recognize characters — real computer vision, done locally on your CPU.

Which language should I choose?

The main language of the document. For mixed English/French documents, English usually handles French accents acceptably — but picking French gives better results.

Can I OCR handwriting?

No — Tesseract is trained on print. Neat block capitals sometimes work; cursive handwriting will not.

How do I get a Word file from a scan?

Run OCR here to get text, then paste it into Word. For PDFs that already have a text layer, PDF to Word converts directly.

Keep going