You have a photograph of a printed page - a notice, a circular, a page of a book, a screenshot of a document - and you need the words out of it as text you can actually edit. Retyping it is the obvious answer and the wrong one. Optical character recognition (OCR) reads the letters out of the picture for you.
What OCR is doing
An image file has no idea it contains words. It is a grid of coloured dots. OCR recognises the shapes of characters in that grid and reconstructs the text they spell. Modern OCR is very good on clean printed text and much weaker on handwriting, heavy stylisation, or a photo taken at an angle in poor light.
How to convert an image to text
- Open the Image to Text tool. It is free, and no account is needed.
- Add your JPG, PNG, WEBP, BMP or TIFF - or up to 30 of them at once.
- Choose the language. English and Hindi both work out of the box; several other Indian scripts are available too.
- Copy the text out.
Getting good results
- Shoot straight on. A page photographed at an angle produces slanted lines that confuse character recognition badly. Hold the phone parallel to the page.
- Get light on it. Even, bright, indirect light. Your own shadow across the page is the single most common cause of a bad result.
- Fill the frame. More pixels per character means better recognition. Crop out the desk.
- Pick the right language. Running Hindi text through English OCR produces nonsense, not a partial result.
When you want a PDF back instead of plain text
If what you actually want is a scanned PDF you can search and select text inside - rather than loose text - use PDF to Searchable PDF, which runs the same OCR but writes an invisible text layer back into the PDF so it stays a PDF. And if you need the content as an editable document, the PDF to Word converter does OCR on scanned pages as part of the conversion.
A note on accuracy
No OCR is perfect. Check numbers especially - dates, amounts, roll numbers, account numbers - before relying on the output for anything that matters. A 1 read as a 7 is a small error in a paragraph and a serious one in a bank statement.