Remove Special Characters

What counts as a special character?

In this tool, a special character is anything that is not a letter, a digit, or a space. That covers the obvious symbols like @ # $ % and parentheses, but also the characters you cannot see or easily type: curly quotes pasted from Word, non-breaking spaces from web pages, zero-width characters hiding inside copied text, bullet points from PDF lists, and emoji.

The default mode keeps letters, numbers, spaces, and line breaks, and removes everything else. Letters here means letters in any language, not just A to Z: accented characters like é and ü, Cyrillic, Greek, Chinese, Japanese, and Korean all survive the cleanup. If your definition of special is narrower or wider than that, the two custom fields under the mode switch let you adjust it without writing a regular expression.

Why Unicode makes this harder than it looks

Old cleanup scripts only worried about ASCII: 95 printable characters, so stripping symbols was simple. Modern text is Unicode, with well over one hundred thousand characters, and that changes the job in two ways.

First, the same visual character can exist in several forms. A straight quote and a curly quote are different code points, so a filter that only knows ASCII leaves the curly ones behind, and they are usually the ones causing trouble. Second, a naive letters means A to Z rule destroys international text: café becomes caf, and every Chinese character disappears from a bilingual document.

This special character remover uses Unicode-aware matching instead. The default mode recognizes letters and digits from any script, removes symbols, punctuation, and control characters, and leaves real words intact whatever language they are in. That is usually what people actually mean when they ask to remove the weird characters.

How to remove special characters from text

1. Paste your text into the box above. The character count updates as you paste.
2. Choose what survives. The default keeps letters, numbers, and spaces in any language. Add characters like .,-@ to the Also keep field, or switch to Remove only mode and list the exact characters to strip.
3. Click Remove Special Characters and copy the result. The counter tells you exactly how many characters were removed.

Before and after

Here is a product blurb straight from a marketing page, complete with curly quotes, an em dash, bullets, a trademark sign, and an emoji:

“Smart” Kettle — 50% off!!! ✨
• Boils in 90 seconds™
• Auto shut-off

Smart Kettle  50 off 
 Boils in 90 seconds
 Auto shutoff

Every symbol is gone and every word and number survives. Notice one detail: shut-off became shutoff, because a hyphen is not a letter, a digit, or a space. If your text relies on hyphens, add - to the Also keep field and they will be protected, which is exactly what the next section is about.

Keep exactly the characters you need

The Also keep field extends the default mode. Whatever you type there is treated as safe: add .,@- when your text contains emails, decimal prices, or hyphenated words, and sales@example.com or 1,299.99 comes through untouched. Add /: if you are cleaning text that contains URLs or timestamps. Order does not matter and each character only needs to appear once.

The second mode flips the logic. Remove only these characters leaves your text alone except for the exact characters you list: paste ™©• into the field and only trademark signs, copyright signs, and bullets disappear. This surgical mode is handy when a document is basically fine and just one or two recurring symbols need to go. In both fields, characters that mean something special in regular expressions are escaped for you automatically.

Where this cleanup shows up

Before you strip everything

A quick sanity check saves rework. Emails need @ and ., URLs need / : ., prices need . and sometimes ,, and phone numbers often need + and -. If any of those appear in your text, put the characters they depend on into the Also keep field before you run the cleanup.

Accented letters are letters here, so résumé stays résumé. If a downstream system insists on pure ASCII, what you need is transliteration, turning é into e, and that is a different job than removal. And if all you want to strip is punctuation while keeping contractions readable, the dedicated punctuation remover gives you finer control.

Choose the narrowest rule that solves the problem

Start with Remove only these characters when you know the exact troublemakers. Removing a recurring bullet, trademark sign, or zero-width character is safer than rebuilding the text from an allowed set. Use the broader Keep letters, numbers, and spaces mode when the destination has a strict format, such as an identifier column that must contain plain words and digits. Then add back only the separators the destination accepts.

Test the rule on a representative sample before cleaning a full export. Include a name with an accent, a non-Latin word, an email address, a price, a date, and any code or identifier your workflow uses. This tool keeps Unicode letters and their combining marks, so languages that build visible letters from several code points remain readable. It does not convert one spelling form into another, transliterate scripts, or guarantee that two visually identical strings have the same underlying encoding.

Removal is not the same as validation or security

Deleting symbols can make a column easier to import, but it does not prove that the result is valid. A cleaned email without its at-sign is not a valid email, a cleaned URL may no longer be a URL, and an order number may lose a meaningful separator. Validate the result against the rules of the system that will receive it, and keep the original value when traceability matters.

This page is also not an HTML, SQL, or script sanitizer. Do not use it as the only protection before inserting untrusted text into a website, database, command, or template. Those systems need context-aware escaping and validation provided by their own libraries. Use TextCleanly for visible text cleanup, not as a security boundary. A practical sequence is: preserve the source, clean a sample, compare the before and after counts, inspect unusual records, and only then apply the same rule to the rest.

FAQ