Image to Text converter — extract text from any image

What OCR does

OCR — optical character recognition — reads the letters in a picture and turns them into real text you can copy, search and edit. Point it at a photo of a document, a screenshot, a scanned page or a sign, and the words come back as text instead of pixels.

Extract text from a screenshot or photo

Screenshots of chats, error messages, slides and web pages convert almost perfectly, because the text is sharp and evenly lit — dark-mode screenshots included, which are inverted automatically before reading. Phone photos of paper work well too when the page is flat, in focus and photographed from directly above. Upload JPG, PNG, iPhone HEIC, WebP, TIFF, BMP or GIF images, or paste a screenshot straight from your clipboard with Ctrl+V. A PDF goes through PDF to Text, or PDF OCR when it is a scan.

Handwriting to text

Switch the image mode to Handwriting for notes, forms and letters written by hand. Neat block capitals and clear joined-up writing are recognised well; rushed scribbles less so. "Flag low-confidence words" highlights anything the reader was unsure of, so you know exactly what to check.

Urdu, Arabic, Hindi, Chinese and 35 more

Auto-detect works out the script — Latin, Arabic, Devanagari, Chinese, Cyrillic and others — and reads with the right languages for it. Setting the language explicitly still makes a real difference for right-to-left scripts such as Urdu and Arabic, and for Chinese. When an image mixes two languages, such as Urdu with English, choose the second one under "Second language".

What happens to the image before it is read

A photo sideways is turned upright, a slightly crooked page is straightened, shadows and uneven light are evened out, small text is enlarged, and light-on-dark text is inverted. The result screen lists whatever was fixed, so you know why a difficult picture still came out right.

Getting the most accurate result

• Fill the frame with the text, and keep the page flat and evenly lit
• Avoid shadows across the page and glare on glossy paper
• Keep "Enhance contrast" and "Correct page skew" on for phone photos
• Higher resolution helps — small or blurry text is the most common cause of mistakes
• Not right first time? Read it again with another language or image mode — it is free
• On a phone, "Take a photo" opens the camera straight from this page; on a computer, paste a screenshot with Ctrl+V or the "Paste image" button

Plain text, Word, searchable PDF, Markdown or JSON

Plain text is for copying and pasting. Word (.docx) is for editing the text as a document, with uncertain words highlighted for you to check. A searchable PDF keeps the image exactly as it looks but adds an invisible text layer, so you can search and select the words — ideal for archiving scanned paperwork. Markdown suits notes apps, and JSON gives developers every word with its position on the page and its confidence score.

Edit, clean up and check the text on the page

The text comes back in an editor, not a dead block. Fix a word, join broken lines into paragraphs, heal words split across lines, remove extra spaces, change the case, then copy it or download it as it now reads. Words the reader was unsure of are highlighted, and the confidence map shows the same thing over your picture, so checking a long page takes seconds.

Read just part of an image

Turn on "Select an area to read" and drag a box around what you need — one paragraph of a screenshot, the total on a receipt, a single formula. Everything outside the box is ignored, so there is nothing to delete afterwards.

QR codes, barcodes, tables and formulas

A QR code or barcode in the picture is decoded and shown beside the text, with its link ready to open. When the image is a table, one click sends it to Image to Excel to keep the rows and columns. And "Math formula (LaTeX)" turns a printed equation into LaTeX you can copy, or paste into Word as a real equation.

Private mode: read it on your own device

For a document that should not leave your phone or computer, open "Read on this device". The text is recognised by your own browser, so nothing is uploaded and no credits are used. It is a little less accurate than the main reader, which cleans the image up first, but it is there when privacy matters most. Your recent results can also be kept in this browser, and only there, to come back to later.

Several images at once

Add up to ten images in one go — the pages of a handout, a set of screenshots, a stack of receipts. Each is read separately and shown on its own tab, and "Combine a batch into one file" puts them all into a single Word document, PDF or text file in the order you uploaded them.

One image, or a whole scanned document

This page is built around a single picture: a photo, a screenshot or one scanned page. When what you have is a multi-page scan, PDF OCR is the tool for it — it reads every page and hands back the same PDF with a text layer underneath, rather than a wall of loose text.