OCR a PDF — free, in your browser
Scanned a document and got a PDF with no selectable text? Run OCR right here — the recognized text downloads as a .txt file. No upload, no account.
What we found testing it
We OCR'd a 3-page scanned lease. English recognition was accurate on clean print; a slightly skewed page had a few swapped characters, as expected. The first run downloaded the engine + English pack (a few MB), then later runs were faster. A Chinese-language scan worked after switching the language dropdown to Chinese.
How to use it
- Open the OCR PDF tool above and choose your scanned PDF.
- Pick the language of the document (English by default; Spanish, French, German, Portuguese, Chinese and Japanese are available).
- Click OCR PDF → text. The first run fetches the OCR engine (a few MB).
- When it finishes, a .txt file with the recognized text downloads.
What OCR can and can't do
| OCR handles | OCR struggles with |
|---|---|
| Clean printed text in a supported language | Heavy handwriting or ornate fonts |
| Multi-page scans, page by page | Very low-resolution or blurry scans |
| Output you can copy, search and paste into Word | Perfect layout reconstruction (it returns text, not a formatted document) |
OCR returns plain text, not a searchable PDF. If you need the text embedded back into the PDF as a layer, that's a different (heavier) workflow — but for 'I just need the words out of this scan', text output is the fastest private path.
FAQ
Is PDFzen OCR free and private?
Yes. OCR runs in your browser with Tesseract.js; the PDF is never uploaded.
What do I get — a searchable PDF or text?
A .txt file with the recognized text. That's deliberate: it's the simplest, most private way to get words out of a scan.
Which languages are supported?
English, Spanish, French, German, Portuguese, Chinese (simplified) and Japanese in this tool.
Why is the first run slow?
It downloads the OCR engine and the language pack (a few MB) once; later runs are quicker and work offline.
My scan came out with errors — why?
OCR accuracy depends on scan quality and language. Use clean, high-resolution prints and pick the right language; expect to fix a few characters on poor scans.
Can OCR read a normal (text) PDF?
It can, but if your PDF already has selectable text, use the plain Extract Text tool — it's instant and exact.