Pull the text out of a photo or scan — English & Hindi
To copy text from a photo of a book page, a notice or a bill, use OCR. You can pick English, Hindi or both. The language pack downloads the first time, so the first run takes a little longer — after that it is quick. Copy the text out or download it as a .txt file.
OCR turns a picture of words into words you can select, search and edit. It works well on printed text that is reasonably sharp, straight and well lit — a book page, a printed form, a screenshot, a clean photocopy.
It struggles with handwriting, with heavily stylised fonts, with faded thermal-paper bills, and with photographs taken at an angle in poor light. Expect to read through and fix a few words. Nothing available today, at any price, is perfect on a difficult page.
Hindi recognition is supported and works reasonably on clean printed Devanagari — a newspaper column, a printed government notice. Accuracy is lower than for English, because conjunct characters and the matra marks above and below the line are genuinely harder to separate. Check the output carefully before you rely on it.
A PDF made from a Word file or a website already contains real text, and pulling it out needs no recognition at all. PDF to Text extracts it exactly, with no mistakes. Use OCR only when the PDF is a scan — a picture of pages, where selecting text with your mouse selects nothing.
Yes, Hindi (Devanagari) is supported and works well on clear, straight scans.
OCR is never 100%. Handwriting and blurry photos produce mistakes — always read the text through once.
Straighten it with the Document Scanner first — accuracy improves a lot.
No. Recognition runs inside your browser. The language data is downloaded to you; your image goes nowhere.
The recognition engine and the language data have to download once, which is a few megabytes. After that it is cached and later runs start immediately.
Not reliably. It is built for printed text. Neat block capitals sometimes work; ordinary cursive handwriting does not.
It reads the words but does not preserve the table structure. For tables inside a PDF, PDF to Excel does a better job.
After the language data has downloaded once, largely yes — but load the page while you still have a connection.
English and Hindi, including pages that mix the two.
🔒 Everything on this page happens inside your browser. Your file is never uploaded to a server.