What OCR solves
A scanned PDF is just pictures of pages. You can read it, but you cannot search for a word, copy a sentence or have it read aloud. OCR — optical character recognition — reads the letters in the pictures and adds real text to the PDF, so the document becomes searchable and copyable.
What happens when you run OCR
Choose the document's language: English, Hindi, Kannada, Tamil, Telugu, Marathi, Bengali, Spanish, French, German or Arabic. Each page is drawn at high resolution and read on your device, and a new PDF is built in which every page keeps its scanned look with a hidden text layer behind it. Up to 100 pages can be processed at a time; long documents take several minutes on slower devices.
Which scans give the best results
Straight, sharp, well-lit scans of printed text work best. Handwriting, very small print, stamps over text, faint photocopies and pages with heavy background patterns lower the accuracy. Straighten tilted scans first with Deskew Scanned PDF.
Common problems and how to fix them
Search finds nothing: check that the right language was chosen and run OCR again. Copied text has mistakes: OCR is never perfect — proofread important parts. The PDF has more than 100 pages: split it with Split PDF, run OCR on each part and merge them again.
Good habits
Run OCR on scans you will search later, keep the original scan, and use Image to Word or PDF to Word on the result when you need to edit the text.