Image to Text (OCR)
Extract the text from a screenshot, photo, or scanned document with OCR — in 18 languages, entirely in your browser. Nothing is uploaded.
Prefer the terminal? See the CLI equivalent →About the Image to Text (OCR)
- Runs entirely in your browser
- 18 languages
- Drop, choose, or paste an image
- Read the whole image or a selected region
- Edit, copy, or download the text
- Instant results
- 100% private — runs in your browser
- Free, no sign-up
Image to Text pulls the words out of a picture — a screenshot, a photo of a page, a receipt, a sign, a scanned document — and gives you selectable, copyable text. It uses OCR (optical character recognition) that runs completely inside your browser with WebAssembly, so your image is never uploaded to a server.
That matters because the images you most want to read text from are often personal or confidential — a bank statement, an ID, a contract, a private message. Every "image to text" website asks you to upload the picture; this one does not. The recognition engine and the language models are served from this site and run locally, and after the first use they are cached so it works offline.
Drop an image, click to choose one, or just paste a screenshot from your clipboard. Pick the language that matches the text for the best accuracy, then edit the result if needed and copy it or download it as a .txt file.
How to use it
- Drop an image onto the box, click to choose a file, or paste a screenshot straight from your clipboard.
- Pick the language of the text — the right model makes recognition much more accurate.
- Wait a moment while it reads (the first run downloads the engine, then it is cached).
- To read just one part, turn on “Select area” and drag a box around it — handy for a single line, a column, or one field on a form.
- Fix any mistakes in the editable result, then copy it or download it as a .txt file.
Good to know
- Nothing uploaded
- The OCR engine, the WebAssembly core, and the language models are all served from this site and run in your browser — no server, no upload.
- 18 languages
- English, Spanish, French, German, Italian, Portuguese, Dutch, Polish, Hungarian, Russian, Ukrainian, Turkish, Czech, Romanian, Chinese, Japanese, Korean, and Arabic.
- Best on clear text
- Printed text and screenshots work great; handwriting and low-quality, skewed, or low-contrast photos are less reliable.
- Region select
- Turn on “Select area” and drag a box to OCR only that part of the image — often more accurate than the whole page, especially with a matching layout mode (single line, single word, or sparse text).
- Cached after first use
- The engine and each language model download once and are cached, so repeat use is fast and works offline.
Frequently asked questions
Is my image uploaded anywhere?+
No. The recognition runs entirely in your browser using WebAssembly, and even the OCR engine and language models are served from this site rather than a third party. Nothing about your image — or the text it contains — leaves your device.
Which languages are supported?+
Eighteen: English, Spanish, French, German, Italian, Portuguese, Dutch, Polish, Hungarian, Russian, Ukrainian, Turkish, Czech, Romanian, Chinese (Simplified), Japanese, Korean, and Arabic. Choose the one matching your image for the best accuracy.
Why is the first scan slow?+
The first time you use it, the OCR engine (about 3 MB) and the language model download to your browser and are cached. After that, scans are fast — usually under a second per image — and work offline.
The text has mistakes — why?+
OCR accuracy depends on image quality. Clear, straight, high-contrast printed text reads almost perfectly; blurry photos, unusual fonts, skew, and especially handwriting are harder. Choosing the correct language helps a lot, and the result is editable so you can fix any errors.
Can I read just one part of the image?+
Yes. Turn on “Select area” and drag a box around the part you want — a single line, a table column, one field on a form — then “Read selection”. It often beats reading the whole page, because Tesseract isn’t distracted by surrounding text; pairing it with the “Single line” or “Single word” layout mode sharpens the result further. “Read whole image” clears the box.
Can it read handwriting?+
Not reliably. This engine is trained for printed text. Neat block capitals sometimes work, but cursive and casual handwriting generally do not.
Related tools
Case Converter
Convert text between camelCase, PascalCase, snake_case, CONSTANT_CASE, kebab-case, Title Case, and more — all at once, in your browser.
Diff Viewer
Compare two texts or code snippets side by side with line-level difference highlighting.
Markdown Editor
Write Markdown with a live GitHub-flavored preview, a table of contents, and one-click export to Markdown or HTML — entirely in your browser, nothing uploaded.
Mojibake Fixer
Repair garbled text where UTF-8 was misread as Windows-1252 — turn “café” back into “café” and “’” back into “’”. Runs entirely in your browser.
Regex Railroad Diagram
Visualize a regular expression as a railroad diagram and read a plain-English, step-by-step explanation. Hover to link the two. Runs in your browser.
Regex Sample Generator
Generate example strings that match a regular expression, then send them straight into the Regex Tester. Every sample is verified to match — all in your browser.