Image to Word

Extract text from a JPG or PNG using OCR. Edit the text before exporting a DOCX and choose whether to include the original image.

This file is decoded in your browser and is not uploaded.

The first time a picture is read, recognition data for the writing on it is downloaded. The file itself is not uploaded.

When the words are still locked in a photo

A notice photographed, instructions screenshotted, a scan saved as JPG: the words are still pixels. This page reads them into a Word file you can edit.

  • Open a JPG or PNG up to 10 MB. Receipts, whiteboards, chat screenshots, and single photos you do not want to wrap in a PDF stay on this device and are not uploaded. Over the size limit, the page stops. Reduce the image and open it again.

  • The page looks at the writing and picks a language model for it. Chinese, English, and other common scripts each have their own data. It is not limited to one language. The first time a script is used, the model downloads and the wait is longer. That download is the model, not your picture.

  • A contract or a slide deck that is already a PDF should not be exploded into JPGs first. That file belongs on PDF to Word, which keeps the paper size and the original positions. This page is for a file that is already a picture.

Correct the reading, then decide whether the photo stays

What was recognized is text. The original picture is still a picture. Wrong words sit in the text box until you change them.

  • Fix a broken line or a wrong name in the text box before you download. The DOCX stores the wording you corrected, not the first draft the recognizer produced.

  • Text only gives you paragraphs you can keep editing, without the image. Use that once you have already checked the wording and only want the words.

  • Including the original picture leaves the image beside the text. Checking amounts and dates against a scan is easier than flipping windows. Names, amounts, and dates are the spots most worth a human look.

  • If you still want to rewrite or restyle, continue in Word or WPS after the download. This page gets the writing out of the picture and lets you correct a draft first.

How to use it

  1. Open a JPG or PNG up to 10 MB. A phone photo of a page, a screenshot, or a scan saved as an image all belong here.
  2. Wait for recognition. The first time a script is seen, its language model downloads, so the wait is longer.
  3. Fix broken lines, names, and numbers in the text box, then download the text alone or with the original picture in the DOCX.

Limits

  • JPG or PNG, up to 10 MB. The picture stays on this device and is not uploaded.
  • Recognition data follows the writing on the picture. Chinese, English, and other common scripts each use their own language model.
  • Download the text alone, or place the original picture in the Word file as well. Correct the text box before you download.

FAQ

Is the picture uploaded?

The picture stays on this device. The first time a script is recognized, the matching language model may download. That download is the model, not your picture.

Can I fix a word it got wrong?

Yes. The words land in the text box first. Correct them, then download the DOCX. The sentences in Word are the version you edited.

Related tools