Corrupted Text Generator

Bury your words under layers of combining marks so they look like a file that went bad.

Plain text
Corrupted text
Runs locally in your browser

About the Corrupted Text Generator

Corruption here is built from combining diacritics, the invisible-on-their-own code points that a font stacks onto whichever character came before them. Piling ten or fifteen of them onto a single letter makes the glyph sprout tendrils that bleed into the lines above and below, which is exactly the look of a corrupted save file or a horror movie subtitle.

The damage selector controls how many marks land on each character: Light adds one or two, Destroyed adds up to eighteen. Spread decides whether those marks climb above the baseline, hang below it, or do both while also weaving strike-through and enclosing marks between the letters. Turn on the look-alike swap and each letter is first replaced by a visually similar character from another script, so the base text itself looks wrong before the marks are even applied.

Spaces, tabs and line breaks are deliberately left clean. Marking a space produces a floating cluster with nothing to attach to, and it breaks word wrapping badly in chat apps. The decoration is also deterministic: the same input always produces the same output, so you can rebuild a caption later without hunting for the exact version you posted.

For a lighter, more rhythmic effect use the Glitch Text Generator, or the Cursed Text Generator when you want odd letterforms rather than raw noise.

How to use

  1. Paste the line you want to wreck.
  2. Raise or lower Damage until the density looks right.
  3. Change Spread if the marks are covering the line above or below too much.
  4. Copy the corrupted text and paste it wherever plain Unicode is allowed.

Common questions

Why does the corrupted text overlap the lines around it?
Combining marks have no width of their own, so a tall stack draws outside its own line box. Lower the damage level or restrict the spread to one direction.
Does the same sentence always corrupt the same way?
Yes. The mark placement is seeded from the text itself, so repeating the same input reproduces the identical result.
Will Discord or Twitter strip the marks?
Discord renders them but clamps very tall stacks. Some platforms limit consecutive combining marks, so heavy output can arrive trimmed.
Can I get the original sentence back?
Not from this page. Deleting every character in the U+0300 to U+036F range from the output recovers the base letters if you need them.