What OCR adds to a scanned PDF
A scan can look like text while actually containing only page images. OCR analyzes those images and adds an invisible searchable text layer to a new PDF. PDFora keeps the rendered page image visible and places recognized text behind it, so the document can be searched while preserving the scanned appearance.
Accuracy depends on the source
- Sharp, straight scans with good contrast are easier to recognize than blurry, skewed, shadowed, or very low-resolution pages.
- Choose the language that best matches the document. Mixed-language pages can reduce recognition accuracy.
- Handwriting, decorative fonts, mathematical notation, tables, and tightly packed forms are harder than ordinary printed paragraphs.
- OCR can misread names, dates, account numbers, totals, and punctuation, so critical information must be checked against the visible page.
What this OCR is best for
Use OCR to make scanned letters, reports, receipts, and archived pages searchable or easier to copy from. It is not a certified transcription service and should not be treated as automatically accurate for legal, medical, financial, or identity records.
Why the first run can take longer
The OCR engine and selected language data may need to be downloaded by the browser before recognition starts. Processing then uses device CPU and memory, so a long scan can take noticeably longer on a phone or low-memory computer. Keep the tab open until the result is ready.
OCR turns page images into recognized text, not guaranteed truth
Optical character recognition estimates characters from pixels. Accuracy depends on scan resolution, contrast, language, font, page rotation, compression artifacts, handwriting, and background noise. Clean printed text can work well, while faint receipts, decorative fonts, stamps, or skewed camera images may produce mistakes.
Always verify names, account numbers, dates, amounts, reference codes, and other critical values against the original page. OCR output should not be treated as automatically accurate enough for legal, financial, medical, or identity-sensitive decisions.
Improve the source before recognition
Straighten pages, use sufficient resolution, improve contrast, and avoid shadows or fingers covering text. If only a few pages need recognition, testing them first can reveal whether the source quality is suitable before processing the entire document.
Language and typography affect recognition
OCR performs best when the recognition language matches the printed text and the page uses ordinary, well-separated characters. Mixed languages, decorative typefaces, vertical text, formulas, handwriting, and low-contrast stamps can reduce accuracy. If the resulting text will be searched later, test a few distinctive names or phrases to confirm that recognition is useful.
Keep the original scan even after OCR. The image remains the reference when recognized text is ambiguous.