Remove Special Characters

About the Remove Special Characters Tool

Symbols and special characters get filtered out, keeping only letters plus whichever categories, spaces, line breaks, numbers, or basic punctuation, you choose to preserve. Useful for sanitizing usernames, cleaning up copy-pasted content, or prepping text for systems that only accept plain alphanumeric input.

Only plain ASCII letters are kept by default; accented and non-Latin letters are removed entirely

The one part of the allowed character set that is never optional is a-zA-Z, the 26 unaccented English letters in both cases. No checkbox extends that range, which means an accented letter like é, ñ, or ü is treated exactly like a symbol and deleted outright, not converted to its plain equivalent: "café" becomes "caf," not "cafe." The effect is much larger on text written in a non-Latin script: every letter in Russian, Arabic, Hindi, Chinese, or any other language whose alphabet is not a-z falls outside this tool's allowed set entirely, so running a sentence in one of those languages through it strips out essentially all of the actual content, leaving only whatever numbers, spaces, or punctuation you chose to preserve. If you want accented Latin letters kept as their plain equivalent rather than deleted, run your text through this site's Remove Accents/Diacritics tool first, then through this one.

Keep basic punctuation covers exactly four characters, not punctuation in general

That checkbox adds only a period, comma, exclamation point, and question mark to the allowed set. Apostrophes, quotation marks, colons, semicolons, hyphens, and parentheses are not included, so even with that option checked, a contraction like "don't" loses its apostrophe and comes out as "dont," and a hyphenated word like "well-known" comes out as "wellknown." If you need a broader set of punctuation preserved, you can easily type those characters into the custom allowed characters field above.

Checkbox order does not matter here, unlike some of this site's other tools

The allowed-character string is built by appending each enabled category's characters one after another, but because the result is used inside a regex character class, [...], the order those characters appear in has no effect on what gets matched; a character class is an unordered set, not a sequence. That is a genuine difference from tools on this site where processing order changes the result, like the ampersand-first rule on the HTML Encoder; here, whichever order the checkboxes happen to be checked in, or whichever order their characters get concatenated in, produces the exact same allowed set and the exact same output.

The regex is assembled safely from fixed fragments and escaped custom characters

Building a regular expression at runtime from a string is normally something to be careful with, since an unescaped regex pattern can behave unpredictably or throw syntax errors. That risk is eliminated here: every checkbox corresponds to a fixed character range, and any custom characters you type are strictly escaped before compiling the regular expression. This guarantees consistent, predictable sanitization across all inputs.

Sixteen possible baseline combinations, one per checkbox state

With four independent checkboxes, there are sixteen possible baseline allowed-character combinations the tool can build, ranging from the strictest (only letters, every other checkbox off) to permissive modes (letters, numbers, basic punctuation, spaces, and line breaks all preserved). Because letters are always kept and you can optionally whitelist custom characters, you can configure the exact output specifications required by your target database, filename schema, or processing API.