How to use the fancy text generator
- Type or paste your text in the box. Every style updates as you type.
- Find a style you like and click the row. The styled text is copied to your clipboard and the row shows “Copied”.
- Paste it into a bio, display name, caption, comment or message.
- To undo the styling, paste fancy text into the Convert back box under the list. It turns bold, italic, script, fraktur, double-struck, circled, squared and fullwidth letters back into normal letters and removes strikethrough and underline marks.
Letters A–Z and a–z are converted in every style. Digits are converted where Unicode has a matching set (bold, double-struck, sans-serif, sans-serif bold, monospace, circled, fullwidth and a few others). Anything without a styled version, such as most punctuation, accented letters and emoji, is left as it is, so your text never loses characters.
Why this is not really a font
Apps such as Instagram and X do not let you choose a font for your bio, so a fancy text generator does not change the font at all. It swaps each ordinary letter for a different Unicode character that happens to look styled. Unicode includes these characters mainly for mathematics: physicists write bold 𝐯 for vectors and mathematicians write ℝ for the real numbers and 𝔤 for a Lie algebra. Because they are separate characters, they survive copy and paste anywhere that plain text is accepted.
The letter A is code point U+0041. Its bold version 𝐀 is U+1D400, a completely different character. That has three practical consequences:
- Search engines and in-app search usually do not match
𝐇𝐞𝐥𝐥𝐨when someone searches for “Hello”. - Most styled letters sit outside the Basic Multilingual Plane, so each one takes two UTF-16 code units. Apps that count UTF-16 units (JavaScript’s
.lengthdoes) count each styled letter as two, which matters for character limits. - Screen readers may read the characters out by their names, or skip them. See the accessibility section below.
The styles and where they come from
| Style | Example | Unicode block |
|---|---|---|
| Bold, italic, bold italic | 𝐀𝐁𝐂 𝐴𝐵𝐶 𝑨𝑩𝑪 | Mathematical Alphanumeric Symbols (U+1D400–U+1D7FF) |
| Script, bold script | 𝒜ℬ𝒞 𝓐𝓑𝓒 | Mathematical Alphanumeric + Letterlike Symbols |
| Fraktur (gothic), bold fraktur | 𝔄𝔅ℭ 𝕬𝕭𝕮 | Mathematical Alphanumeric + Letterlike Symbols |
| Double-struck | 𝔸𝔹ℂ 𝟙𝟚𝟛 | Mathematical Alphanumeric + Letterlike Symbols |
| Sans-serif, sans bold, sans italic, sans bold italic | 𝖠𝖡𝖢 𝗔𝗕𝗖 𝘈𝘉𝘊 𝘼𝘽𝘾 | Mathematical Alphanumeric Symbols |
| Monospace | 𝙰𝙱𝙲 𝟷𝟸𝟹 | Mathematical Alphanumeric Symbols |
| Circled (bubble) | ⒶⒷⒸ ⓐⓑⓒ ①②③ | Enclosed Alphanumerics (U+2460–U+24FF) |
| Black bubble, squared, black squared | 🅐🅑🅒 🄰🄱🄲 🅰🅱🅲 | Enclosed Alphanumeric Supplement (U+1F100–U+1F1FF) |
| Fullwidth | ABC abc | Halfwidth and Fullwidth Forms (U+FF00–U+FFEF) |
| Small caps | ᴀʙᴄ | Phonetic Extensions, IPA Extensions, Latin Extended-D |
| Upside-down | ɐqɔ | IPA and other Latin letters that look rotated |
| Strikethrough, underline, slash | A̶B̶C̶ A̲B̲C̲ | Combining Diacritical Marks (U+0300–U+036F) |
Black bubble and squared letters only exist in capitals, so lowercase input is shown in capitals for those styles. Small caps and superscript are borrowed from phonetic alphabets, and a few letters have no small-cap or superscript form: there is no small-cap x and no superscript q, so those stay as normal letters. Small-cap q uses ǫ, the closest common lookalike.
The “missing” letters in script, fraktur and double-struck
If you build a script alphabet by simply adding an offset to A, some letters come out as empty boxes. That is because the Mathematical Alphanumeric Symbols block deliberately leaves holes: script B, fraktur C, double-struck R and a few others were already encoded years earlier in the Letterlike Symbols block, so Unicode reserved the matching slots instead of encoding them twice. A correct generator has to jump to the older character for those letters. This page does:
| Style | Letters taken from Letterlike Symbols |
|---|---|
| Script | ℬ (B, U+212C), ℰ (E, U+2130), ℱ (F, U+2131), ℋ (H, U+210B), ℐ (I, U+2110), ℒ (L, U+2112), ℳ (M, U+2133), ℛ (R, U+211B), ℯ (e, U+212F), ℊ (g, U+210A), ℴ (o, U+2134) |
| Fraktur | ℭ (C, U+212D), ℌ (H, U+210C), ℑ (I, U+2111), ℜ (R, U+211C), ℨ (Z, U+2128) |
| Double-struck | ℂ (C, U+2102), ℍ (H, U+210D), ℕ (N, U+2115), ℙ (P, U+2119), ℚ (Q, U+211A), ℝ (R, U+211D), ℤ (Z, U+2124) |
| Italic | ℎ (h, U+210E, the Planck constant) |
Because these letters come from a different block, they can look slightly different from their neighbours in some fonts. That is a font design issue, not a mistake in the text.
Accessibility: fancy text and screen readers
Styled Unicode letters are not letters with formatting; they are symbols. Screen readers handle them inconsistently. Some read 𝐇𝐞𝐥𝐥𝐨 as “mathematical bold capital H, mathematical bold small e…”, some skip the characters, and some read them correctly. Upside-down text is read as a list of unrelated symbols, and strikethrough text may announce a “combining long stroke overlay” after every letter.
Practical guidance:
- Use fancy text for a word or two of decoration, not for whole sentences or anything people need to understand.
- Never put essential information (prices, dates, links, contact details) in styled characters.
- Keep hashtags in plain text.
#𝐬𝐮𝐦𝐦𝐞𝐫is not the same tag as#summer. - On your own website, use CSS (
font-weight,font-style,font-family) instead. The text stays real text and assistive technology reads it normally.
Where fancy text works, and where it doesn’t
Anywhere that accepts general Unicode text will keep the characters: bios, captions, comments, chat messages, document editors and most display names. Whether they look right depends on the fonts on the reader’s device. Current phones and computers generally include fonts for all of them, but older devices sometimes show an empty box (“tofu”) for the Enclosed Alphanumeric Supplement characters (black bubble and squared letters), which are the newest of these sets.
Some fields deliberately reject them. Usernames and handles on most platforms only allow ASCII letters, digits, dots and underscores, and email addresses and domain names have their own rules. Some apps also normalise the text when you save it, which can turn styled letters back into normal ones.
Troubleshooting
I see boxes or question marks
The device viewing the text has no font containing those characters. Choose a style from the Mathematical Alphanumeric or Enclosed Alphanumerics blocks (bold, italic, script, circled), which have the widest font support.
Strikethrough looks broken or misaligned
Combining marks are drawn over the previous character, and their exact position depends on the font. Some fonts draw the stroke slightly too short, leaving gaps between letters. By default the mark is also added after spaces so the line is continuous. Tick Keep combining marks off spaces if you only want the words themselves struck through.
My character counter says the text is too long
Most styled letters take two UTF-16 code units (and four bytes in UTF-8), and combining styles add one extra character per letter. The counter above the list shows code points and UTF-16 units for your input so you can compare.
Search or copy-paste into another tool gives strange results
Paste the text into the Convert back box. It applies Unicode NFKC normalisation, which maps the mathematical, enclosed and fullwidth letters to their plain equivalents, then removes the combining strike and underline marks. Small caps and upside-down text have no official link to the normal letters, so they are mapped back with a lookup table.
Do the conversion in code
The bold alphabet is a straight offset from ASCII, so a few lines of JavaScript reproduce it:
const bold = s => [...s].map(c => {
const n = c.codePointAt(0);
if (n >= 65 && n <= 90) return String.fromCodePoint(0x1D400 + n - 65); // A-Z
if (n >= 97 && n <= 122) return String.fromCodePoint(0x1D41A + n - 97); // a-z
if (n >= 48 && n <= 57) return String.fromCodePoint(0x1D7CE + n - 48); // 0-9
return c;
}).join("");
bold("Hi 5"); // "𝐇𝐢 𝟓"
"𝐇𝐢 𝟓".normalize("NFKC"); // "Hi 5"
Python can undo the styling the same way:
import unicodedata
unicodedata.normalize("NFKC", "𝓗𝓮𝓵𝓵𝓸 ℍ𝕚") # 'Hello Hi'
Script, fraktur, double-struck and italic need the exception table above; a plain offset produces unassigned code points for those letters. For a longer walkthrough, see the fancy text generator guide.
Frequently asked questions
How do I copy and paste fancy text?
Type your text, then click the row with the style you want. It is copied to the clipboard straight away, and you can paste it anywhere with Ctrl+V or Cmd+V, or by long-pressing on a phone.
Does fancy text work on Instagram, TikTok and Discord?
Yes in bios, captions, comments and messages, because the styles are ordinary Unicode characters. Usernames and handles usually only allow plain letters, numbers, dots and underscores, so styled letters are rejected there.
Why do some letters show as boxes?
The device reading the text has no font with that character. Bold, italic, script and circled letters are supported almost everywhere; the black bubble and squared letters are newer and fail most often on old devices.
Is fancy text bad for accessibility?
It can be. Screen readers may read each styled letter by its Unicode name, such as “mathematical bold capital H”, or skip it. Use it sparingly for decoration and keep important information in normal text.
Can I turn fancy text back into normal text?
Yes. Paste it in the Convert back box on this page. Unicode NFKC normalisation restores bold, italic, script, fraktur, double-struck, circled, squared and fullwidth letters, and a lookup table handles small caps and upside-down text.
Why are the script B, E and R different from the other script letters?
They were encoded in the older Letterlike Symbols block (ℬ U+212C, ℰ U+2130, ℛ U+211B), so the matching slots in the Mathematical Alphanumeric block are reserved and empty. A correct generator uses the older characters, which can look slightly different in some fonts.
Does fancy text hurt search or hashtags?
Usually yes. Styled letters are different characters, so search engines and hashtag systems treat #𝐬𝐮𝐦𝐦𝐞𝐫 and #summer as different words. Keep hashtags and keywords in plain text.