About TextToHandAI

TextToHandAI is a free text to handwriting converter that runs in your browser. You paste typed text, pick a pen and a sheet of paper, and it draws the words out as handwriting you can download. That is the whole product.

It is built and maintained by Ayush Nautiyal. It is not a company. It is one small tool, maintained in public, with a changelog you can read.

Where it came from

A friend of mine spent a weekend copying out forty pages of typed revision notes because her tutor only accepted handwritten submissions. Forty pages. Her wrist hurt for days.

So I went looking for a tool that could do it for her, and most of what I found fell into two groups. Some were a cursive font pasted onto a white rectangle, which fools nobody, because every letter a on the page is pixel-identical. The rest did something more convincing but put the download button behind a $9/month plan, or stamped a watermark across the middle of the page.

I wrote the first version over about three weeks in early 2026. It has grown since then, mostly from emails people send asking for a script their language actually uses.

How the rendering works

Your text is laid out line by line against the page geometry you chose, then every character is drawn individually onto an HTML canvas. Each glyph gets its own small offsets:

• a tilt of a fraction of a degree, in either direction
• a baseline that sits a pixel or two above or below where it “should”
• slightly uneven spacing before the next character
• a change in opacity, which reads as the pen pressing harder or lighter

On top of that, each line gets a very shallow sine wave so it drifts the way a line of real writing drifts away from a rule.

The numbers are not random in the usual sense. They come from a seeded generator keyed to the page, line and character position, which means the same text always produces the same page. Your preview and your exported PDF match exactly. Nothing shifts under you when you change the ink colour.

Arabic, Thai and Devanagari get different treatment. Those scripts join and stack, so drawing them glyph by glyph breaks them. For those fonts the tool jitters whole words instead, which keeps the shaping intact at the cost of a little variation.

What it does not do

It is worth being clear about the limits.

It cannot learn your handwriting. There is no way to upload a sample and have the tool imitate it. That is the single most requested feature and it is not available, because doing it properly needs a trained model and a server, and a server is the thing this tool deliberately does not have.

It cannot read handwriting either. That is OCR, which is the opposite direction. Google Lens and Apple’s Live Text both do it well and for free.

It will not make a page that passes as yours if a teacher knows your handwriting. People sometimes ask about this. The honest answer is that the output looks handwritten, not like your hand, and if the rule is “write it yourself” then a converter does not satisfy it. There is more on that in the guide about handwritten assignment rules.

Privacy, concretely

Layout, drawing and export all happen in your browser through the Canvas API. There is no API call carrying your text anywhere, because there is no backend to call. You can open your browser’s network tab, paste a whole essay, export a PDF, and watch nothing go out.

The only thing kept on your device is a texttohandai-theme key for light or dark mode and a list of fonts you starred. Both live in localStorage. Details are in the privacy policy.

Two things do load from third parties: the handwriting faces come from Google Fonts, and the site uses Google Analytics for page counts. Neither one sees what you type.

Getting in touch

Bug reports are genuinely useful, especially ones that name the browser and include a screenshot. Font requests get read too, and several scripts are on the site because somebody asked. The address is on the contact page.

If you want to see what the tool can do before anything else, the converter is on the front page and the font gallery lists every style.