Make a scanned PDF searchable

Choose pages and a confidence threshold for the source

Make the one-page synthetic scan searchable

The scanned-example.pdf sample contains one landscape A4 image-only page based on ocr-example.png. Its visible English phrases let you test the added layer directly.

1. Choose scanned-example.pdf and select Load PDF pages. Set Pages for OCR to all.
2. Choose Minimum word confidence 30% and select Make PDF searchable.
3. Download the result, open it in a PDF viewer and search for BROWSER OCR SAMPLE. Check the reference digits 2048 and amount 1250 against the visible scan.
4. Copy a recognized line into a text editor and compare every word with the page. Confirm the output still has one page and the visible image remains in place.

Expected result. The sample keeps its visible page and gains an invisible layer for accepted words. Searching or copying tests that layer; merely seeing Searchable PDF ready does not establish that every phrase was recognized correctly.

Check the downloaded result

• Compare PDF pages, OCR pages and Searchable words in the result. A selected page can contribute no accepted words even if other pages make the overall job succeed.
• Search several distinct words and numbers, then copy text into an editor. Verify reading order and punctuation, not just that one keyword can be found.
• Check selection highlights against the scanned text. Word placement is approximate and does not reconstruct the original fonts, tables or exact reading order.
• Compare visible pages and final file size with the source. Adding a text layer can increase bytes and does not automatically make the document accessible or editable like a word-processing file.

Troubleshooting

No confident English words were found

No recognized word survived the confidence, character and position checks. The source may be unclear or unsuitable for the English model.

Try this: Use a clearer upright scan or try a lower confidence threshold. Check the actual recognized output rather than lowering the threshold solely to obtain a successful status.

A name or symbol is absent from copied text

The added text layer currently retains printable ASCII characters only. Accents, non-Latin scripts and some currency or punctuation characters can be discarded even if visible in the scan.

Try this: Compare the hidden text with the page and use a suitable multilingual OCR workflow when those characters matter. Do not treat the copied text as a complete transcript.

Copied words are duplicated or search highlights are confusing

OCR was added to a page that already had text, or the document was processed repeatedly. Existing text layers are not removed by this operation.

Try this: Return to the original source and select only the pages that need OCR. Avoid repeatedly adding a new layer to the same already-searchable output.

The page selection exceeds the device limit

OCR is bounded to 5, 10 or 20 selected pages per job depending on the device profile, with additional word and output-size limits.

Try this: Choose a smaller selection using page numbers or ascending ranges, such as 1,3-5. Keep track of processed pages and verify each output before continuing.

Limits and compatibility

Confidence and page rendering

Selected pages render at bounded resolution for English recognition. Average confidence is the mean of page-level engine scores, not a measured accuracy rate or a score for only the accepted words.

Original content

The browser adds text to the existing PDF locally. It does not redact, sanitize, repair form behavior, correct the visible scan or produce tagged accessibility structure. Password-protected sources must be unlocked first.

Your files and form values are processed on this device. Downloaded copies remain wherever you save them. See how to check local processing and our privacy policy for site data and optional services.

Common questions

Will unselected pages be removed?

No. The full document remains. Page selection controls where this run adds recognized text; unselected pages do not gain a new layer.

Should I compress the PDF after adding OCR?

Check the method first. This site's raster-compression path removes selectable text, including an OCR layer. If compression is necessary, prepare a readable compressed source first, then add OCR and recheck final size.

Continue your document workflow

• Recognize one image as plain textUse this when a text draft is sufficient and a searchable PDF is unnecessary.
• Correct sideways pages firstPrepare upright visible text before running recognition on scanned pages.

Maintained by Shakti Prasad Ram. Found a mismatch? Report the tool, browser and steps without sending a sensitive document. Back to the tool