PDF-007
OCR PDF
A scanned PDF is a photograph of a page: you can see the words, but no software can read them. Optical character recognition looks at the shapes and works out the characters, turning the picture back into text you can search, select and copy.
- Free
- No signup
- Private · runs locally
How to ocr pdf
- Upload the scanned PDF.
- Pick the language of the document so the recogniser knows which characters to expect.
- Wait for each page to be processed, then download the searchable copy.
How to tell whether you need OCR
Open the PDF and try to select a line of text. If the cursor highlights individual words, the text layer is already there and you do not need this tool - go straight to PDF to DOCX or PDF to Markdown.
If the selection draws a rectangle over the whole page, or nothing happens at all, the page is an image and OCR is the only way to get text out of it.
What makes recognition accurate
OCR output always needs proofreading. Numbers are the usual weak spot, and a misread digit in an invoice or a reference code is easy to miss because the surrounding text looks perfect.
| Factor | Effect on accuracy |
|---|---|
| Scan resolution | 300 DPI or higher is reliable; below 200 DPI errors climb sharply |
| Contrast | Clean black on white is best; grey or yellowed paper confuses edges |
| Straightness | Skewed or curved pages from phone photos lose accuracy |
| Font | Standard print is accurate; handwriting is not supported |
| Language setting | A mismatch produces confident nonsense, not an error |
Frequently asked questions
Is my PDF uploaded to a server?
No. The file is opened and rewritten by your browser, so it never leaves your device. That matters for contracts, invoices and anything with personal data in it.
Does it handle handwriting?
No. The engine is trained on printed type. Handwritten notes will produce unusable output.
Why is it slow on the first run?
The language model downloads once and is then cached by your browser. The first page is the slow one; the rest are much faster.
Can I trust the recognised text?
Treat it as a good draft. Accuracy on a clean 300 DPI scan is high but never perfect, so proofread anything that matters - especially figures.