Start with these steps
- Open OCR PDF and choose one PDF from your device.
- Select up to three printed languages, primary language first, then start recognition.
- Review uncertain words, prepare a searchable PDF or UTF-8 text result, and download it before leaving the tool.
A searchable layer keeps the scan visible
A scanned page stores a picture of its words. OCR attempts to recognize those words and add invisible text over the original page positions. Searching and copying use the recognized layer; the visible scan stays unchanged.
Correcting recognized words changes the search and copy layer, not the visible page. OCR PDF does not reconstruct editable Word tables or add PDF accessibility tags automatically.
Choose languages that match the print
The supported languages are English, Spanish, Portuguese, French, German, Italian, Indonesian, Hindi and Marathi. Select up to three, with the primary printed language first. For a mixed Marathi and English form, select Marathi, then English.
OCR is recognition, not translation. Automatic language detection and deskew are not included. Use orientation settings for sideways scans, and choose a page range when only part of the document needs recognition.
Review meaningful details before saving
Compare names, dates, amounts, accents and identifiers against the original preview. Low-confidence flags help prioritize review; they are not measured accuracy percentages. Handwriting and damaged scans may remain inaccurate even after retrying.
PDF output includes every original page. UTF-8 text output contains processed pages only. Existing text is retained, and failed or unprocessed pages stay visually unchanged in the PDF. Download from the result page before refreshing or closing the tool; recognition and results are held in browser memory.
Allow for device limits
The tool processes documents in the browser without sending the PDF to a conversion server. Software and selected language data load from this website. The upload screen shows a 200 MB limit on supported desktops, or 50 MB on constrained devices.
Large or complex scans still depend on available memory and processing time. Keep the tab open, use a smaller page range when useful, and stop recognition with the common loader if necessary. Completed pages remain available for review.
Check the finished file
- Select the printed languages, not the language you want as a translation.
- Check names, identifiers, dates and amounts against the visible original.
- Check search and copy in the downloaded PDF.
- Download before refreshing or closing the tool.
Tools for this workflow
Read the tool’s current limits before choosing a file. See our privacy policy or contact support for help.