The pages are never transmitted anywhere; recognition happens on your machine, and the tool keeps working with the network disconnected once the app has loaded. That matters for the scans that most need it — contracts, identity documents, records. Accuracy follows the scan: clean, straight pages recognize well, while skewed or faded pages benefit from Clean scanned PDF beforehand (before OCR, since cleanup rebuilds pages as images). Choose plain-text output if you only need the words in a .txt file. Spot-check names and figures against the page image, which is left exactly as it was scanned.
How it works
- Open Bindery in your Windows browser and drop the scan into OCR
- Pick the language and searchable-PDF or plain-text output
- Download and test with Ctrl+F in Edge
Questions people ask
Will Windows Search index the text afterwards?
The output is a standard PDF with a text layer, which Windows Search indexes when a PDF filter is installed — Edge and Acrobat Reader both provide one — and Ctrl+F finds words in any viewer. The page images are unchanged, so the scan looks exactly as it did.
Does OCR work offline on my PC?
Yes. The engine and language packs are part of the app and load from Bindery, not a cloud service. Once the app has loaded you can disconnect entirely and recognition continues — a straightforward way to confirm that nothing is being uploaded.
Which languages can it read?
Eleven: English, Spanish, French, German, Portuguese, Italian, Arabic, Simplified Chinese, Japanese, Russian and Hindi. Select the document’s language before running, because the engine uses the language pack to resolve ambiguous letter shapes. The wrong pack costs accuracy, and each pack loads from Bindery the first time it is used.