Cursive Text Generator

Type once, get both script alphabets, and see exactly which of your characters Unicode cannot make cursive — before you paste it into a bio and find out there.

The eleven holes, and why every other cursive tool falls into them

The plain script alphabet lives at U+1D49C to U+1D4CF inside Unicode's Mathematical Alphanumeric Symbols block. It looks like a tidy run of fifty-two letters, so the obvious way to build a cursive generator is arithmetic: take the letter's position in the alphabet, add it to the base codepoint, emit the result. That works for forty-one letters and produces garbage for eleven.

Eight capitals — B, E, F, H, I, L, M and R — and three lowercase letters — e, g and o — were already encoded years earlier in the Letterlike Symbols block, where mathematicians had been using them for Hamiltonians, Laplace transforms and power sets. Unicode does not encode the same character twice, so when the mathematical alphabets were added the eleven duplicate slots were left permanently unassigned. Arithmetic still happily computes them. The result is a codepoint no font has ever drawn, which your reader sees as a rectangle, a question mark, or nothing at all, depending on their device.

This page shows you the damage instead of hiding it. The coverage list names every character it could not style and every one it patched, and the dropdown lets you pick the trade-off yourself. Patching is the sensible default, but it is not free: ℬ, ℰ, ℱ, ℋ, ℐ, ℒ, ℳ, ℛ, ℯ, ℊ and ℴ come from a different block with a different coverage story, so a font that has the mathematical alphabet may lack the letterlike one and produce a word where one capital is visibly heavier, differently slanted, or a box after all. If your text contains several of them, the second panel is the safer choice: Mathematical Bold Script has no holes at all, all fifty-two letters present, because none of them was ever encoded anywhere else.

What you are actually pasting

This is not a font. Nothing here changes how your text is drawn; it changes which characters your text is made of. 𝒜 is not an A in a cursive typeface, it is a separate character with its own identity, name and codepoint. That is why the appearance survives a copy and paste into a field that offers no formatting at all, and it is also the source of every problem below.

Script letters live above U+FFFF, so each one costs two UTF-16 code units. Anything counting your text in code units — which is what a plain .length gives you in most languages — sees a twelve-letter cursive name as twenty-four characters. Platforms differ on whether their limits are counted in code units, codepoints or grapheme clusters, so the tool prints both numbers under each panel and you should trust the field over either of them.

The patched letters make the accessibility cost worse here than in the other stylers. A patched string is drawn from two blocks at once, so a screen reader can read part of a word and abandon the rest, and the eleven substitutes are the characters least likely to be in its pronunciation table. What screen readers do with styled characters generally, and where styling is still safe, is set out in why styled text breaks screen readers. Keep a plain version of anything that has to be read — and keep the searchable spelling of your name somewhere in the account, because platform search matches codepoints, not shapes.

It cannot see the device you are pasting into

Everything runs in this page. Your text is never sent anywhere, there is no request when you type, and the tool keeps working with the network off. What it cannot do is tell you how a specific device will render the result, because that depends on fonts installed on someone else's machine that this page has no way to inspect. Treat the two panels as a comparison rather than a preview, send yourself a test message before committing, and remember that a bio which looks elegant to you may be a row of boxes to a reader on an older phone.

When the letters do not show up

Why do some letters show up as empty boxes even after patching?

Because the patch moves the problem rather than removing it. The eleven substitutes are real, assigned characters, but they sit in Letterlike Symbols rather than the mathematical block, and font coverage is decided per block. Older Android builds and some Windows font stacks ship one and not the other. If you see a box in the first panel, switch to bold script, which draws from a single unbroken range.

Why can I not make numbers or accented letters cursive?

Unicode encodes no script digits whatsoever. The Mathematical Alphanumeric Symbols block does contain digits — bold, double-struck, sans-serif, sans-serif bold and monospace — but the script and bold script alphabets end at the letters. There is no cursive 7 in existence to paste. Accented letters have the same answer for the same reason: the mathematical alphabets cover unaccented A–Z and a–z only, so é, ñ and ü have no script form, and neither does any Greek, Cyrillic or CJK character. Those are passed through untouched and listed in the coverage report.

Should I use script or bold script?

Bold script, in most cases. It draws from a single unbroken range, so it has none of the eleven holes and needs no patching, which makes it far more likely to render the same way on a device you have not tested. Plain script is the lighter, more elegant of the two on a machine that has the fonts for it — use it when you can check the result where it will actually be seen, and use bold script when you cannot.

Will people still find me if my display name is cursive?

Not by typing your name. Search on almost every platform matches the characters stored, not the shapes they resemble, so a script string is a different string from the plain-letter version of it and will not come up for it. If people need to find you by name, keep the plain spelling in the handle, the bio, or a pinned post, and treat the styled version as decoration on top of a findable account.

Can I use this in a username instead of a bio?

Usually not. Bios, display names, captions and post bodies accept arbitrary Unicode on most platforms. Handles and account names generally do not — they are commonly restricted to ASCII letters, digits, underscores and periods, so the field either rejects the input or quietly strips it. The reliable move is to paste into the real field and save, rather than trusting any generator's promise about a platform.