OCR (Optical Character Recognition) recognises letters and digits in an image and turns them into text. It is useful when a document is a photograph or a scan and its words cannot be copied by selecting them.
When OCR is useful
First try selecting and copying a few lines from the original file. If a PDF already has a correct text layer, extracting that text may be preferable to recognising the page again. An image or a scanned PDF needs OCR.
Preparing the document
Use the clearest original available: a straight page without cropped lines, glare or strong shadows. Enlarging a small image does not restore missing detail. Select the language that matches the document; an incorrect language can make similar letters harder to recognise.
Choosing an output
For ordinary text, choose an available text format in the converter. When rows and columns matter, use a table recognition conversion. A text output does not by itself guarantee preservation of the original layout, pictures or positioning of blocks.
Checking the downloaded result
Compare names, dates, identifiers and amounts with the original. Pay particular attention to similar characters such as O and 0, number separators and small print. A clear scan helps, but neither resolution nor engine selection guarantees an error-free result. Handwritten notes, complex layouts and damaged pages need extra review.