Searchable PDF / OCR
Run OCR on a scanned PDF free, entirely in your browser, using an on-device OCR engine — convert scanned pages into searchable, selectable text.
- Add
- Adjust
- Make
- Keep
How to use this tool
- Add a scanned or image-based PDF using the file picker or drag-and-drop — no need to check whether it is 'real' text first.
- Choose which pages to OCR and the language of the text on those pages. The first time you pick a language, its pack downloads on-device with visible progress; after that it works fully offline.
- Pick a rendering resolution: 300 DPI is the recommended default for printed text, 600 DPI helps with very small print, and 150/72 DPI are faster drafts. Keep deskew, grayscale, and contrast-boost on for typical scans, and turn them off for already-clean documents to save time.
- Set a confidence threshold (60% by default). After the run, the summary shows how many recognized words fell below it, so you can review them in a follow-up pass.
- Download the result: your original pages are visually unchanged, but now carry an invisible searchable text layer you can select, copy, and find within.
About this tool
Run OCR on a scanned PDF free, entirely in your browser, using an on-device OCR engine — convert scanned pages into searchable, selectable text. Nothing is uploaded to a server; the searchable PDF is generated locally and downloads straight from your device.
Privacy & limitations
OCR accuracy depends on scan quality, typeface legibility, and language-pack coverage; handwriting, very low-contrast scans, and highly decorated pages are recognized poorly. The invisible text overlay is best-effort alignment, not pixel-perfect — text selection and copy may snap to boxes that are slightly offset, especially near margins and in multi-column layouts. OCR runs entirely on your device, so very large or many-page documents can take a long time, and memory use scales with page size and DPI (a 600-DPI letter page is roughly 8 MB of pixel data per page). For non-Latin scripts such as Devanagari (Hindi) the text layer comes from the OCR engine's own Unicode page, which is stretched to the page box; on rotated pages, or if that page cannot be embedded, those words fall back to the Latin-only standard font and are skipped, so they will not be searchable or selectable. Pages that carry a rotation in the PDF itself (for example a landscape page inside a portrait document) get an overlay aligned to the rotated view, where placement is approximate.
Last updated:
Frequently asked questions
How do I convert a scanned PDF into editable text?
Add the scanned PDF and run OCR — the tool recognizes text on each page locally using on-device OCR and adds an invisible, selectable text layer on top of the original scan, so the document looks the same but text becomes searchable and copyable.
Does OCR work on handwritten text?
OCR here is tuned for printed text. Handwriting recognition accuracy is much lower, and handwriting is not a supported use case for this tool. Neat block capitals sometimes come through, but cursive generally does not, so expect to retype handwritten pages by hand.
Which languages does the OCR support?
You can choose English, Hindi, Spanish, Portuguese or French. The first time you pick a language, its data downloads once and shows progress, then OCR works offline. Your pages are not uploaded. Pick the language that matches the text on the page for the best accuracy.
Why is OCR slow or inaccurate on some pages?
OCR runs on your device, so long documents and high resolutions take time and memory. Accuracy depends on scan quality, so low-contrast, skewed or decorated pages are recognised poorly. The default 300 DPI suits printed text; try 600 DPI for very small print.