Remove Punctuation

Delete punctuation marks and symbols from text.

Tidy up the leftover spaces

Input

Output

Characters: 0Words: 0

Stripping Punctuation for Clean Text

This tool removes every character that isn't a letter, number, or whitespace using Unicode-aware matching, so commas, periods, quotes, brackets, hashes, and symbols all disappear while accented letters and non-Latin scripts are preserved. It's handy for tokenizing text, preparing data for analysis, or cleaning input before indexing — all done privately in your browser.

Built and maintained by Meet Shah · Last updated 5 August 2026

What this tool is used for

Stripping punctuation before a word-frequency analysis.
Cleaning text for a comparison where punctuation should not matter.
Preparing a phrase for a palindrome or an anagram check.
Normalising a value before using it as a key.
Removing symbols from OCR output before proofreading.

Frequently Asked Questions

What counts as punctuation?: This uses the Unicode punctuation categories (P*), so it removes more than ASCII: em dashes, curly quotes, ellipsis characters, and CJK punctuation like 。and 、. A naive [^a-zA-Z0-9 ] filter would also strip accented letters, which is usually wrong.
Will it break contractions and hyphenated words?: Yes, by design — "don't" becomes "dont" and "well-known" becomes "wellknown". If you need the word boundary preserved, replace hyphens with spaces before stripping rather than after.
Why remove punctuation at all?: It is a standard normalisation step before text analysis: word frequency, n-gram extraction, search indexing and machine-learning tokenisation all typically strip punctuation so that "word" and "word," count as the same token.
Are currency and maths symbols removed?: No. $ and + are Unicode Symbol (S*) categories, not Punctuation, so they survive. That distinction is deliberate in the standard and is why "punctuation" removal does not flatten arithmetic expressions.
What about apostrophes in names?: They are removed too, so O'Brien becomes OBrien. If you are normalising names rather than analysing prose, exclude the apostrophe explicitly — losing it changes the name rather than cleaning it.
What happens to apostrophes inside words?: They go, so `don't` becomes `dont` and `O'Brien` becomes `OBrien`. That is usually wrong for names and contractions, which is why a strip intended for text analysis normally keeps the apostrophe and removes everything else.
Are typographic quotes and dashes covered?: Only if the rule is Unicode-aware. An ASCII punctuation class misses curly quotes, en and em dashes and the ellipsis character that word processors insert automatically — which is exactly the punctuation pasted text contains.
Should hyphens be removed from compound words?: It depends what follows. For word counting, removing the hyphen splits `well-known` into two words and changes the total; for search indexing that is often what you want. The right answer comes from the consumer, not the text.

Common errors and gotchas

Removing hyphens and apostrophes, which changes words rather than tidying them.
Losing sentence boundaries, which any later per-sentence analysis then cannot recover.
Stripping punctuation from code, where it is syntax rather than decoration.
Assuming only ASCII punctuation is removed, when other scripts have their own marks.
Leaving the whitespace punctuation used to occupy, which produces doubled spaces.