A scan is essentially a picture — text you can see but not select. OCR (optical character recognition) turns it into a genuinely readable, searchable text layer.
Step 1: pick the right language
In OCR, the recognition language must match the document: Chinese docs tick Chinese; for mixed Chinese/English, tick both, or you'll get dropped characters or garbled text. This single setting drives accuracy most.
Step 2: enhance, then recognise
A skewed phone shot, uneven lighting or low contrast all hurt accuracy. Run Scan Cleanup or Scan Enhance first to denoise, boost contrast and brighten, then OCR — the result is usually much better.
Step 3: save as a searchable PDF
After recognition, merge the picture and the text layer into a "searchable PDF" — you see the original image yet can select and search the text. If the source is images, Image to PDF first, then recognise.
When OCR still struggles
- Handwriting or decorative fonts: low machine accuracy, proofread manually;
- Very complex layout (multi-column, dense tables): recognise in segments;
- Severely blurry or incomplete: enhance first; if that fails, re-scan.
OCR runs locally in your browser; files are never uploaded. Always proofread important data, especially amounts and IDs.
📖Related reading
Make scanned PDFs clearer: deskew, remove black edges, enhance
Scanned or photographed PDFs fail in predictable ways: skewed pages, black borders, grey backgrounds, faint text. A fixed order works far better than repeatedly clicking enhance. Here is the four-step flow.
OCR & scanFrom paper to searchable PDF: a clean scanning workflow
Photos or scans of paper aren't usable as-is: wrong orientation, grey backgrounds, and no text you can search. The standard flow is Image to PDF → straighten & brighten → OCR to make it searchable → optimize the size. Each step has a matching tool, and it all runs locally.
OCR & scanTurn paper into a searchable PDF: scan, crop and OCR in one flow
How do paper contracts, invoices and notes become searchable, copyable PDFs? The flow: images to PDF → crop black edges → straighten → local OCR. Everything runs in your browser.
🧭Related solutions
Image to PDF: turn photos and scans into a booklet
Turn phone photos and scans into one PDF: enhance, deskew and clean the background first, then assemble. Bitonal docs become smaller and sharper as pure 1-bit.
ConvertScan OCR and text recognition
Image PDFs have no text layer — you can't search or edit them. Enhance, deskew and clean first, then OCR into a searchable, copyable text layer, or convert straight to Word.
ConvertPDF to Word: get an editable document
Want to edit a PDF? Text PDFs convert straight to Word/text; scans need OCR first. Also PDF to Markdown and Word to PDF — all local.