Read handwriting from a picture
Separated, hand-printed letters — a form filled in by hand, block capitals, a neatly lettered note — can often be read. Joined-up writing cannot, by this or any other general OCR tool, and this page will tell you which you have rather than guess.
- Never stored
- No queue, no waiting
- No signup, no watermark
How it works
Add your picture
A photograph or scan of the handwriting. The straighter and better lit, the more comes back.
See how it did
The reader's own confidence is shown beside the text, because with handwriting that number is the useful part. Low means it struggled, and the text below should be read as a draft.
Copy what is worth keeping
Take the text away and correct it, which is usually far quicker than typing it all again.
What can be read, and what genuinely cannot
Joined-up handwriting is not recognised here, and the page tells you so rather than returning plausible nonsense. That is the most important sentence on this page. Reading print works by finding the boundaries between letters and matching each shape — and in cursive there are no boundaries, because the letters are one continuous stroke whose shapes change depending on what they connect to. It is a fundamentally different problem, not a harder version of the same one.
What does work, often well, is separated hand-printed letters: block capitals, a form filled in one character per box, a shopping list written carefully, engineering or architectural lettering. Those are discrete shapes with gaps between them, which is the same problem as print with more variation in it. Form boxes are the best case there is, because the boxes do the separating for you.
Accuracy on handwriting is far below accuracy on print in every case, so the result is a draft to correct rather than an answer to trust. You are shown how confident the reading was, which is there to be used: a low confidence on a digit in an account number matters far more than a low confidence in the middle of a sentence you can read from context. The genuinely dangerous errors are the confident ones — a 1 read as a 7, a 5 as an S — because nothing flags them and both are plausible.
Getting a better result is mostly about the picture. Flat, evenly lit, photographed straight on rather than at an angle, in focus, and as large as your camera will manage. Dark ink on plain unlined paper reads better than pencil on squared paper, because the lines of a grid are strokes too. The output is plain text — no layout, no colours, and no indication of what was crossed out.
When you want something else
For genuine cursive, the honest answer is that a hosted service will read what this cannot. Google Cloud Vision, Azure AI Vision and Amazon Textract all handle joined-up handwriting to a standard nothing running locally approaches, because they are trained on vastly more material. That means sending your pages to a third party, which is exactly the trade this site exists to avoid — a reasonable choice for a recipe card, a poor one for medical notes or a diary.
For historical documents and archives specifically, Transkribus is built for the job and is used by real archives: it can be trained on a particular hand, which is what makes centuries-old manuscripts readable at all. And for a page or two, typing it out is frequently faster than correcting a poor recognition — worth measuring honestly before committing to the second approach.
Frequently asked questions
Will it read my handwriting?
If the letters are separated and roughly upright — block capitals, a form filled in by hand, careful printing — often yes, and the result is worth correcting rather than retyping. If your writing is joined up, almost certainly not. That is not a limitation of this page; general OCR tools are trained on printed type and recognise letter shapes, and cursive has no separate letter shapes to recognise.
Then why does this page exist?
Because "can OCR read handwriting" is a real question with a real answer, and most tools that claim to do it are doing exactly the same general OCR as this one without saying so. Hand-printed forms are genuinely common and genuinely readable. Showing you the confidence, and being plain about cursive, is more use than a confident wrong answer.
What gets the best result?
Good light with no shadow across the page, the camera square on rather than at an angle, the writing filling the frame, and dark ink on plain paper. Ruled or squared paper is fine. Pencil is harder than pen because it is grey rather than black.
Is my picture stored anywhere?
No. The language pack comes from this site the first time you use a language; the picture itself is never stored and never sent anywhere, so it works with the Wi-Fi off.
Can it read handwriting in other languages?
The same 120-odd languages are available, but the caveat is bigger, not smaller: the readers are trained on printed type in each script, so hand-printed Bengali or Arabic is harder again than hand-printed English. Try it and look at the confidence.
Is there something that reads cursive properly?
Handwriting recognition that copes with joined-up writing needs to be trained on handwriting specifically, and the good ones run on a server rather than in a browser tab. We would rather point you at that honestly than pretend a printed-text reader does it.
Good to know: Joined-up handwriting is not recognised — this is a printed-type reader and the page says so rather than returning plausible nonsense. Hand-printed letters, block capitals and filled-in forms often work. Accuracy on handwriting is far below accuracy on print in every case, so treat the result as a draft to correct.
Put this tool on your website
Free for any blog, class page or help article. Paste one snippet and your visitors can use it right on your page.