Zalgo Text Remover

How to use it

1. Paste the glitchy text into the box, or press Insert an example to see how it works.
2. Under What to remove, choose Glitch marks only to keep normal accents, or Every accent too to strip all of them.
3. Read the clean text and the count of marks removed below it.
4. Press Copy clean text.

What the remover does

Zalgo text, also called glitch or cursed text, is normal letters with a pile of Unicode combining marks attached. The letters are still there underneath. Removing the glitch means finding those marks and dropping them.

The tool works in four steps:

1. It rewrites your text in decomposed form (NFD), so every accented letter becomes a base letter followed by its marks.
2. It walks through the text and groups each base letter with the marks that follow it.
3. It drops the marks, or keeps them if the group is small enough to be a real accent (depending on the mode).
4. It rewrites the result in composed form (NFC), so accents that survive are normal single characters again.

The counter under the output tells you how many marks were removed.

The two modes

Glitch marks only

This is the default. A letter with two marks or fewer is left as it is, because ordinary accents look like that. A letter with three or more marks is reduced to the bare letter. Most Latin-script words never need more than two marks on one letter, so French, Spanish, German, Czech and Vietnamese text passes through untouched.

Every accent too

This strips every combining mark, including accents. Use it when you want pure ASCII-style letters, when light glitching is left behind in the default mode, or when you need to remove combining underline and strikethrough.

Worked examples

Take the word Hi and run it through the glitch text generator at intensity 25, above only. In one run it came out as 7 characters: H with 3 marks and i with 2 marks, 5 marks in total.

This is the main thing to know about the default mode. Light glitch text, with only one or two marks per letter, looks like accents to the tool and is kept. The same word at intensity 12 came out as 5 characters with 3 marks, and the default mode changed nothing. If you see leftovers, switch to Every accent too.

Heavy text behaves differently. In another run, Hello at intensity 100 with all three directions was 122 characters, 118 of them marks. The default mode returned exactly Hello.

Now a normal sentence: Crème brûlée, Việt Nam, naïve. In the default mode it is returned unchanged. With Every accent too it becomes Creme brulee, Viet Nam, naive. Note that ệ in Việt carries two marks, a circumflex and a dot below, which is why it still counts as a normal letter in the default mode.

Where it can go wrong

Scripts that rely on combining marks

Thai, Arabic, Hebrew, Hindi and many other scripts write vowels, tones or other parts of the word as combining marks. Every accent too will strip them. The Thai text ที่ นี่ becomes ท น, which is a different, broken word. For these scripts, use Glitch marks only, and check the result. Even then, a cluster with three or more marks gets reduced.

Styled letters are not marks

Bold, italic, script and similar styled letters are separate characters from the Mathematical Alphanumeric Symbols block (U+1D400 to U+1D7FF). They are not combining marks, so this tool leaves them alone. The LinkedIn text formatter has a Clear button that maps them back to plain letters.

Underline and strikethrough

Text with a line through it or under it usually uses a combining mark after each letter, for example U+0336 for strikethrough. That is one mark per letter, so the default mode keeps it. Choose Every accent too.

Other hidden characters

Zero-width spaces, look-alike letters from other alphabets and control characters are not combining marks. If text still looks odd after cleaning, it may contain something of that kind. The remover does not touch them.

Why the marks pile up in the first place

Unicode lets a single base letter carry any number of combining marks. There is no built-in limit, which is what makes zalgo text possible. Most of the marks used for the effect come from the Combining Diacritical Marks block (U+0300 to U+036F, 112 characters). Apps differ in how many they draw. Some clip them, some let them spill over neighboring lines, and some have crashed on them. Wikipedia notes problems with some versions of Apple's Messages app and with Gmail.

If you want to make glitch text rather than clean it, go to the glitch text generator. For other Discord and LinkedIn text tools, see the Discord text formatter and the LinkedIn text formatter. If a string looks like encoded data rather than glitch, Base64 decode may help.

Sources

• Wikipedia: Zalgo text
• Wikipedia: Combining Diacritical Marks
• Wikipedia: Unicode equivalence (NFC and NFD)
• Wikipedia: Mathematical Alphanumeric Symbols

Checked October 2026.