Fancy Text Generator: 27 Unicode Styles to Copy and Paste

Unicode was built to encode every character people write, and that includes the letters mathematicians use. In a formula, a bold 𝐱 and an italic 𝑥 can mean different things, so Unicode gave each its own code point. Version 3.1, published on 16 May 2001, added a block called Mathematical Alphanumeric Symbols for exactly this purpose. It holds complete alphabets in bold, italic, bold italic, script, bold script, fraktur, double-struck, sans-serif and monospace, plus bold, double-struck, sans-serif and monospace digits.

Most of the alphabets sit in neat runs: if bold capital A is U+1D400, bold capital B is U+1D401, and so on to Z. A generator only has to add an offset to each plain letter. The other styles come from blocks made for other jobs: circled letters from Enclosed Alphanumerics, small capitals and superscripts from the phonetic alphabets linguists use, fullwidth letters from the forms used beside Chinese, Japanese and Korean text, and strikethrough or underline from combining marks that attach to the character before them.

Because each style is made of real characters, the result is ordinary text. It can be copied, sent, searched for exactly and saved in any text field. It is also why the same text looks slightly different on different phones: the letters are drawn by whatever font the reading device has for that part of Unicode.

Every style in the generator above, with the same sample so you can compare them, and the honest limit of each. A style with no digits leaves numbers plain; a capitals-only style turns your lowercase letters into capitals. The pages linked in the generator go deeper on the most-used ones.

Here is the detail most generators get wrong. When Unicode added the mathematical alphabets in 2001, some of those letters had already been encoded years earlier in a block called Letterlike Symbols, because they were common in science: ℎ for the Planck constant, ℝ for the real numbers, ℌ for Hilbert space. Unicode does not encode the same character twice, so in the new block those 24 positions were left empty and marked as reserved, with a note pointing to the older character.

A generator that only adds an offset will land on those empty positions. An empty position has no glyph in any font, so the letter turns into a box or a question mark on every device. This generator fills the gaps from Letterlike Symbols instead, which is why the cursive “Hello” above mixes ℋ, ℯ and ℴ with the rest. The table shows the cursive set; the italic, Old English and double-struck pages show theirs.

On 27 October 2026 we read the character map of every system font on a Mac running macOS 26.5.1 (365 font files, leaving out the LastResort font that only draws placeholder boxes) and of the fonts Android lists in its platform font configuration: Roboto, the Noto Sans Symbols subsets, Noto Sans Lisu, Noto Sans CJK and Noto Color Emoji. For every style we checked whether each character it produces is drawn by at least one font.

The result: every character of all 27 styles is drawn on both platforms, and none of the 24 empty positions is drawn by any font on either. So on a current iPhone-class Apple system or a current Android phone, a correct generator's output should display; boxes almost always mean a reserved code point, an older system, an app that bundles its own fonts, or Windows, which we could not measure because its fonts were not on the test machine.

Free-text fields such as bios, captions, posts, comments and most display names accept any Unicode character, so styled text pastes in as long as the app's character limit allows it. Fields that act as identifiers are different. Discord, for example, documents that usernames may only contain the letters a–z, the digits 0–9, underscores and periods, while display names “can be almost anything”. A styled username is rejected there; a styled display name is fine.

Two practical side effects follow from the letters being different characters. First, search: someone searching for “hello” will not find 𝐡𝐞𝐥𝐥𝐨, because the characters do not match. Do not style the words people need to find you by. Second, length: each mathematical letter is one character to Unicode but two code units in UTF-16, the encoding many apps use internally, so some character counters count it twice and a bio fills up faster than you expect. Combining marks such as strikethrough add a character for every letter they decorate.

Assistive technology reads characters, not appearances. A screen reader meeting 𝐇𝐞𝐥𝐥𝐨 may announce each letter's Unicode name (“mathematical bold capital H…”), spell it out, or skip it, depending on the reader and its settings. Either way, a sighted reader sees one word and a blind reader may hear noise.

The W3C's accessibility guidelines ask that meaning and structure never depend on visual presentation alone, and that real text be used rather than imitations of styling. In practice: use the app's own bold or italic when it has one, keep styled text short (a name or a single decorative word rather than a whole post), and never put essential information such as dates, links or instructions in a fancy style.

Styles are not interchangeable. Some read easily at small sizes and some do not, some have digits and some do not, and some carry meanings of their own.

Three different things get called “fonts” online. Knowing which one you are using saves a lot of confusion when something stops working.

Most problems have one of a few causes, and the fix follows from the cause.

None of these alphabets were added for social media. Mathematicians use typeface as meaning: bold for vectors, double-struck ℝ and ℕ for number sets, fraktur for algebraic structures, script for categories and transforms. Before 2001, that distinction lived only in typesetting software. Unicode 3.1 made the letters themselves portable so a formula could be copied between programs without losing the difference between 𝐯 and 𝑣.

Pasting those letters into a profile is a side use the standard never planned for, which is why some combinations are missing: mathematics never needed an italic digit, so Unicode never encoded one. Knowing that history is the quickest way to predict what a style can and cannot do.