Read a scanned page or photo into an editable Word document
Processed in your browser. Nothing is uploaded. Updated on 24 Oct 2026.Image to Word reads the printed text in a picture and writes it into a real .docx you can edit, with the lines joined back into paragraphs. Accuracy is bounded by the reading: a sharp scan is near perfect, handwriting fails, and layout is dropped. Up to 10 pictures, 20 MB each.
Press Select pictures, or drop pictures of text onto the page.
Choose the Language, and under Line breaks pick Join into paragraphs or Keep every line.
Correct any wrong words in The document text box.
Press Download .docx; the reading happens on your device.
Image to Word runs optical character recognition on a picture and writes what it found into a Word document. The output is a genuine .docx: paragraphs of real text that open in Word, Google Docs, Pages and LibreOffice, and can be edited, searched and restyled like anything you typed yourself.
The honest limit of the tool is the recognition, and everything else follows from it. A screenshot or a flatbed scan of printed type is read almost perfectly. A photograph taken square on, filling the frame, in even light, is read well. A photograph that is blurred, tilted, half in shadow or shot with a glare across the page produces text that needs correcting line by line, and at that point retyping may genuinely be faster. Handwriting does not work at all: the recognition is trained on printed type, and a handwritten note comes back as nonsense. None of that is a fault to be fixed in a later version; it is what OCR is, and knowing it before you start is what lets you take a better photograph instead of trusting a bad result.
Layout is not preserved either. Columns, tables, headers, page numbers and pictures are dropped, and what arrives is the words in reading order. The one layout decision the tool does make is worth knowing about: lines are joined back into paragraphs, because a line break in a picture is where the page ran out of width rather than where a sentence ended. A document built one paragraph per line looks fine until you edit it and every line stays where it was. For an address, a form or a menu, where the breaks are the content, keeping every line is the better setting and it is one chip away.
The reader is downloaded to your browser once, about 4 MB, and then everything happens on your own device. No picture is uploaded, nothing is stored, and no account is needed, which is the point when the page is a contract, a payslip or a medical letter. Hindi is shipped alongside English, and documents that mix the two have their own setting.
Accuracy depends entirely on the picture. A screenshot or a flatbed scan of printed type is read almost perfectly; a photograph that is sharp, square on and evenly lit reads well; a blurred, tilted or shadowed one needs correcting line by line. Always check names, numbers and dates against the picture.
No. The recognition is trained on printed type, so a handwritten note comes back as nonsense or as nothing at all. No free tool does handwriting well, and this one says so rather than handing you a document full of invented words that looks like it worked.
No. The reader is downloaded to your browser once, about 4 MB, and after that the picture is read and the Word file written on your own device. Nothing reaches a server and no account is needed, which matters when the page is a payslip, a contract or a medical letter.
No. Columns, tables, headers, page numbers and pictures are dropped, and what you get is the words in reading order as ordinary paragraphs. Rebuilding a table from a photograph means guessing where the cells were, and a table guessed wrong takes longer to fix than one built again.
Because a line break in a picture is where the page ran out of width, not where the writer ended a sentence. Joining them gives paragraphs that reflow when you edit them in Word. For an address, a form or a menu, switch to keeping every line, which makes each line its own paragraph.
Yes. Hindi recognition data is shipped here, which most free converters never load, and there is a setting for documents that mix Hindi with English. Choosing the language the picture is actually in matters more than any other setting: Hindi read as English comes back as gibberish.
Image ko Word banaye in three steps: add the picture, choose the language, then download the .docx once you have checked the text. Photo saaf aur seedhi honi chahiye, warna text galat aata hai. Handwriting kaam nahi karti. Nothing is uploaded; the reading happens in your browser.
Still stuck? Ask about Image to Word and anybody who knows can answer. No account needed.
Same job, other tools: Compress Image to 100 KB – Get any photo under 100 KB for forms and portals · Compress Image to 200 KB – Bring a photo under 200 KB with the detail still there · Resize Image for Instagram – Square, portrait and story sizes in one click · Resize Image for LinkedIn – LinkedIn banner at 1584×396 pixels, set and ready to crop · Image Converter – Convert between JPG, PNG, WebP, AVIF, BMP, TIFF and ICO, from ten input formats.
On the related tools: HEIC to PNG – Convert iPhone HEIC photos to lossless PNG.
Also works as: PNG to Word · Scanned Image to Word.
In the collections Office File Tools · OCR Tools.