You scan a contract, a bank statement or a stack of old letters, and you get a PDF. It looks fine. Then you try to find one word with Ctrl+F, or to copy a paragraph, and nothing happens. The file is a set of pictures, and a picture of a word is not a word. To search it you need a text layer, and OCR is the way to add one.
What a searchable PDF is
A searchable PDF is a normal PDF with two things on each page. The first is the page image, exactly as scanned. The second is an invisible layer of text placed on top of or behind the image, with each recognised word positioned where it appears in the picture. You see the scan, but when you search or select, the PDF viewer uses the hidden text. This is sometimes called a PDF with a text layer, or a sandwich PDF.
This is different from extracting text into a separate file. If you only want the words, Extract text gives you a plain text you can copy or download. If you want to keep the original look and make the PDF itself searchable, use OCR to PDF.
Get a better scan first
OCR is only as good as the picture. A few minutes spent on the scan saves a lot of corrections later.
- Scan at 200 to 300 dpi. Lower resolution blurs small letters, and much higher only makes the file large.
- Keep the page flat and square to the camera. A curved page near the spine of a book bends the lines and confuses the reader.
- Use even light with no shadow across the page, and avoid glare on glossy paper.
- Make sure the page is the right way up. Rotate sideways pages first with Rotate PDF.
- Crop away dark borders if you can, using Crop PDF.
How to make a scanned PDF searchable with DocPro
- Open OCR to PDF and choose the scanned PDF. You can also drop several photos of pages, and they become one PDF in the order you add them.
- Pick the language or languages that appear in the document. Choose every language that is actually present, because more languages make the reading slower.
- For a PDF, optionally enter the pages to process, such as 1-3, 5. Leave it blank for all pages. By default, pages that already contain selectable text are kept as they are and skipped.
- Press Make searchable PDF and wait. Each page is read in turn, and the progress is shown per page. You can cancel at any time.
- Download the new PDF, then try searching for a word you know is in it.
The first run loads the recognition engine and the language data, which are bundled with the page, so nothing is downloaded from another site. After that the work happens on your device. A long document can take a while, especially on a phone.
What to expect
OCR is very good on clean, printed text and weaker on handwriting, stamps, small print, tables and pages with a coloured background. The tool reports an average confidence for the pages it read, but that is only a guide. A word read wrongly is still stored in the text layer, so searching for the right word may miss it. Proofread the parts you depend on, such as names and numbers.
The output keeps the original page images, so the file is usually larger than the scan and can be much larger than a compressed original. If size matters, Compress PDF can reduce it afterwards, with a trade-off: compression turns pages into images again and removes the text layer, so compress first and run OCR last, or keep both versions.
Arabic, Persian, Urdu and Hindi are supported for recognition. For the hidden text layer in these scripts, DocPro embeds a small font on your device. If a character cannot be covered by any bundled font, that word is left out of the text layer, and the result tells you how many words were skipped.
When you need something else
- If you want an editable Word document, try PDF to Word on a PDF that already has text, or extract the text and paste it into your editor.
- If you only have photos of pages, turn them into a PDF first with Images to PDF, or let OCR to PDF do it in one step.
- If you need to hide parts of a scanned page, use Redact PDF before or after OCR. Redacted pages are images with black boxes, so run OCR afterwards if you also want them searchable.
Privacy
Scans often hold the most private pages you own: IDs, medical letters, tax forms. Many online OCR services upload the pages to their servers to read them. DocPro reads them on your device, so the pages are not sent anywhere. You can check that in your browser's network tab while you run it.
Frequently asked questions
Does OCR change how the PDF looks?
No. Each page keeps its original scanned image, and the recognised text is hidden behind it.
Can it read handwriting?
Not reliably. OCR is built for printed text. Handwriting usually comes out wrong or is skipped.