PDF Text Recognition (OCR) — Currently in the Works
A scanned PDF looks like text to the eye but is just a picture inside the file. OCR reads the characters from that picture and adds a searchable, copyable text layer.
This page accepts no file.
This tool isn't available yet
We don't build file pickers for features that can't actually run. Here's the current status and what it would take to ship it.
| Current status | Not available (in preparation) |
|---|---|
| Technology required | Tesseract (Apache-2.0) plus language training data, image pre-processing (deskew, binarize), and a job queue |
| Difficulty | High |
| Estimated server cost | $40/month and up, because CPU time per page is high; heavy use needs extra queue workers |
| External API needed | Optional — a commercial OCR API is worth considering when accuracy matters |
What you can do today
Text recognition in a scanning app
Many phone cameras and document-scanning apps recognise text at capture time.
Get the source document
Where possible, asking for the original electronic file instead of a scan wins on both accuracy and file size.
Some scripts are harder to recognize than Latin text due to combined letters and mixed vertical writing and tables. Offering a low-accuracy result as a "searchable PDF" would actually produce wrong search results, so we plan to introduce it after quality verification.
How are your files handled?
This page accepts no file.
This page has no file picker, so no data is processed. See the Privacy Policy and File Retention Policy for details.