OCR (optical character recognition) turns a picture of text into real text. Most disappointing OCR results come from choosing the wrong route or starting with a poor scan, not from the tool. This guide helps you decide before you upload. The conversion itself is done by OCR PDF.
Do you need OCR at all?
Open the PDF and try to select a line of text with your mouse.
- The text highlights. The file already contains text. You do not need OCR for searching or copying. If you want the data in another format, use a converter such as PDF to Word or PDF to Excel.
- Nothing highlights, or the whole page is selected as one block. The page is a picture. This is where OCR helps.
- Some pages highlight and others do not. The file is mixed. In Automatic mode the existing text stays as it is, and text inside images is recognized too.
Which route to take
| Your goal | Route |
|---|---|
| Keep the PDF but make it searchable and copyable | OCR PDF |
| Get the text from a photo or image file | Image to Text |
| Edit a scanned document in Word | Run OCR PDF first, then PDF to Word. Converting a scan directly gives a Word file with the pages as pictures, not editable text. |
| Get a scanned table into a spreadsheet | PDF to Excel reads scans with OCR automatically. See the PDF to Excel guide. |
| Scan is faded, blurry or crooked | Improve it first with Enhance PDF, then run OCR. |
Limits to plan around
- File size: up to 30 MB per file.
- Pages: up to 10 pages per file. For a longer document, choose the pages you need with a range such as
1-3, 5, or cut the file first with Split PDF. - Processing: OCR runs on our server, and the tool shows when it is waiting for the OCR server. If the server is busy, try again in a few minutes or with fewer pages.
- Password-protected files are not accepted. Remove the password first with Unlock PDF.
- Too large? Select fewer pages, or reduce the file with Compress PDF.
Getting accurate text
- Fix the source, not the settings. A straight, sharp, evenly lit scan beats any option. If a page is crooked, faded or photographed at an angle, rescan or enhance it before recognition.
- Select only the language or languages in the document. Fewer languages give faster and more accurate results.
- Check numbers and dates. OCR can confuse similar characters, such as 0 and O or 1 and 7. Compare important figures with the original.
- Do not expect handwriting to work. OCR is built for printed text. If the tool reports that no text was found, the pages may be empty, handwritten, or in a language that was not selected.
Languages
You can add every language used in the document, up to 4 at once. A large list of languages is available, including Arabic, Hebrew, Chinese, Japanese and Korean, and Latin and Cyrillic scripts. For a bilingual document, such as Arabic and English, select both. For a document in one language, select just that one.
Problems and fixes
- "No text was found". Check that the right language is selected, that the pages are printed rather than handwritten, and that the scan is readable.
- The text is full of mistakes. Improve the scan quality, remove extra languages, and try again.
- "The OCR server did not finish in time". Try fewer pages, or try again later.
- "Too many pages". Select a range of up to 10 pages, or split the file.
- The result is still not editable in Word. Run OCR first, then convert the OCR output.