This is the tool where local processing matters most, because scans are usually the sensitive documents: contracts, identity papers, medical and financial records. With Bindery the pages are never transmitted; recognition happens on your machine, full stop. Accuracy depends on the scan — clean, straight pages read well, skewed or faint ones less so, and Clean scanned PDF can fix modest skew and weak contrast first (run it before OCR, as it rebuilds pages as images). Once the text layer exists, the rest of Bindery opens up: Compare, Ask your PDF, PDF → Excel and PDF → Word all need it.
How it works
- Drop the scan and open OCR; pick the language and searchable-PDF output
- Recognition runs on your device — pages stay pixel-identical
- Download and test with Ctrl+F; spot-check a few names and numbers
Questions people ask
Does OCR change how the scan looks?
No. The searchable output keeps every page image exactly as it was and adds an invisible text layer above it for searching and selecting. If you would rather have only the words, the plain-text option writes a .txt with a marker per page instead.
Which languages can it recognize?
Eleven: English, Spanish, French, German, Portuguese, Italian, Arabic, Simplified Chinese, Japanese, Russian and Hindi. Pick the document’s language before running — the engine uses a language pack to resolve ambiguous shapes, and the wrong one costs accuracy. Each pack is loaded from Bindery itself the first time it is used.
Is the recognized text always correct?
No OCR is. Printed text on a clean, straight scan is recognized very reliably; faxes, handwriting, small type and skewed pages produce errors. Treat the text layer as a search aid and check figures that matter against the page image, which is unchanged.