OCR → Markdown
Turn scanned PDFs, photos, and screenshots into clean Markdown. Recognition runs 100% in your browser — nothing is uploaded, nothing is stored.
DROP A SCAN OR PHOTO
.pdf .png .jpg .webp — up to 5 files, 10 pages per PDF
Runs 100% in your browser · Nothing is uploaded
Your Markdown will appear here. Files are read on your device and never uploaded.
Output filename
Please read: local OCR has limits
This free tool runs a compact recognition model on your own device. It is genuinely private, but it is not magic: low-resolution photos, skewed pages, decorative fonts, dense tables, and handwriting will produce mistakes. Always proofread the Markdown before publishing or feeding it to an AI pipeline.
Need higher accuracy, handwriting support, whole-document PDFs, or large volumes? Contact us about server-side OCR access, which runs more powerful models on dedicated hardware.
Local vs server-side OCR
Local (this page)
- + Maximum privacy — files never leave your device
- + Free, no account, works offline after first load
- + Instant Markdown with headings and lists detected
- − Compact model: struggles with poor scans and handwriting
- − Speed depends on your device; 10 pages per PDF
Server-side (via contact)
- + More powerful models, better accuracy on hard material
- + Handwriting, dense tables, whole documents, batches
- + Same clean Markdown pipeline at the end
- − Files must be uploaded for processing
- − Access is currently granted on request
Tips for accurate results
- 01Scan at 200–300 DPI. Phone photos work, but hold steady and fill the frame with the page.
- 02Pick the document's language before converting — it is the biggest accuracy lever after scan quality.
- 03Straighten skewed pages and boost contrast; gray-on-gray text is the most common failure.
- 04For tables, verify every cell — OCR reads text, it does not understand table structure.
- 05Proofread numbers, names, and codes character by character. OCR confuses 0/O, 1/l, and 5/S.
FAQ
How does the free OCR work?
Your scan or photo is read by an OCR engine running entirely in your browser. The recognized text is then cleaned up — wrapped lines are joined, headings and lists are detected — and handed to you as Markdown. The engine downloads its language data on first use, so you need a connection the first time; after that, files never leave your device.
Is my scanned document private?
Yes. Unlike most online OCR tools, nothing is uploaded to any server — recognition happens on your own device. Close the tab and nothing remains. This makes local OCR a good fit for contracts, medical records, and other sensitive scans.
How accurate is browser-based OCR?
Very good on clean, high-contrast scans at 200–300 DPI with straight, printed text. Expect mistakes on low-resolution photos, skewed pages, decorative fonts, dense tables, and handwriting — always proofread the output. If you need higher accuracy on difficult material, high-volume processing, or handwriting recognition, contact us about server-side OCR access.
Which languages are supported?
English, German, French, Spanish, Dutch, Portuguese, and Italian. Pick the document's language before converting — matching the language is the single biggest accuracy lever after scan quality.
Can I OCR a whole multi-page PDF?
The free in-browser tool processes the first 10 pages of a PDF. For longer documents, split the PDF first or convert it in parts. Larger volumes are exactly what the server-side OCR option is for — get in touch via the contact page.
What is the difference between local and server-side OCR?
Local OCR runs on your device: maximum privacy, free, but limited to one recognition model and your device's speed. Server-side OCR runs larger, more accurate models on dedicated hardware: better results on poor scans, tables, and handwriting, plus whole-document and batch processing — at the cost of uploading the file. Both end in the same clean Markdown pipeline.