Page & Bell

Special Character Remover

This remover strips punctuation, symbols, numbers, emojis, accents, or whitespace from any text using documented Unicode-aware patterns — not the naive [^a-zA-Z] regexes that silently destroy Hindi, Arabic, and Chinese characters. Tick the categories you want gone, or flip to “keep only” mode to whitelist letters, numbers, and spaces. Every active option shows the exact regular expression it runs, and the result updates live with a per-category count of what was removed.

Patterns applied

  • Punctuation: /\p{P}/gu
  • Symbols ($ € ₹ © + = …): /\p{S}/gu

Paste some text above and the cleaned version appears here instantly, with a count of everything that was removed.

How to use the special character remover

  1. Paste your text into the box — usernames, filenames, CSV exports, scraped content, anything.
  2. In Remove mode, tick the categories to delete: punctuation, symbols, numbers, emojis, accents, extra whitespace, or line breaks.
  3. Or switch to Keep only mode and choose a whitelist: letters, letters + numbers, or letters + numbers + spaces.
  4. Check the count badges to confirm what was removed, then copy the result or download it as a .txt file.

Why naive regexes destroy non-English text

The most common snippet on the internet for this job is text.replace(/[^a-zA-Z0-9 ]/g, ""). Run that on नमस्ते दुनिया and you get an empty string — every Devanagari character falls outside a-z, so the “cleaner” deletes the entire sentence. The same happens to São Paulo (the ã vanishes), to Zürich, and to any CJK text. The correct approach uses Unicode property escapes with the u flag: \p{L} matches letters in all 160+ scripts Unicode defines, and \p{M} keeps the combining marks that scripts like Devanagari and Arabic depend on. That is exactly what this tool runs, and the pattern panel shows you each regex so you can reuse it in your own code.

Real cleanup tasks this handles

  • Usernames and handles: Keep only letters + numbers to turn Alex Carter! 🎉 into AlexCarter — emoji, space, and punctuation gone in one pass.
  • Filenames: Remove symbols and punctuation, then enable Strict ASCII if your target system chokes on Unicode — useful before uploading to legacy systems that reject anything outside plain English letters.
  • CSV and SQL import prep: Stray curly quotes, em dashes, and invisible characters inside spreadsheet exports are a classic cause of invalid byte sequence errors on import. Strip symbols and normalize whitespace before loading.
  • Scraped or OCR text: OCR output is littered with misrecognized symbols (¦, ¬, ®). Removing the symbol class while keeping letters and digits rescues the readable content.

Remove punctuation only — a worked example

Ticking just the punctuation box runs /\p{P}/gu. Input: “Hello, world!” — it's a (test) for $500. Output: Hello world its a test for $500. The quotes, comma, exclamation mark, apostrophe, brackets, and even the em dash all disappear, because Unicode classifies the em dash as dash punctuation (category Pd). The $ sign survives: currency signs are symbols (category Sc), removed only when you also tick Symbols. This separation is deliberate — “remove punctuation” and “remove symbols” are different jobs, and lumping them together is how other tools end up deleting the $ from price lists.

Frequently asked questions

Will this tool delete Hindi, Tamil, Arabic, or Chinese text?

Not unless you ask it to. The letter classes use Unicode property escapes — \p{L} matches letters in every script, and \p{M} preserves combining marks like Devanagari matras, so क्षेत्र or العربية survive intact. Only the optional Strict ASCII toggle removes non-Latin scripts, and the label warns you explicitly before it does.

What counts as a symbol versus punctuation?

We follow Unicode general categories. Punctuation (\p{P}) covers full stops, commas, quotes, brackets, hyphens, and the danda (।) used in Hindi. Symbols (\p{S}) covers currency signs like $, €, and ₹, math operators like + and =, © and ™, and arrows. Emojis technically live in the symbol category too, which is why the emoji option runs first — so its count stays accurate.

How are emojis with skin tones and family sequences handled?

As single units. A family emoji is a sequence of several codepoints joined by zero-width joiners (U+200D), a flag is a pair of regional indicator symbols, and a thumbs-up with a skin tone carries a modifier codepoint. The emoji pattern consumes the whole ZWJ sequence, the variation selector, and the modifier together, so you never get leftover invisible joiner characters polluting the output.

What does the accents option actually do to é or ñ?

It decomposes the text to Unicode NFD form, which splits é into e + a combining acute accent (U+0301), then strips every combining mark with \p{M}. The base letter survives: café becomes cafe, ñ becomes n. This is the standard technique for building ASCII-safe slugs and filenames without deleting the letters themselves.

Why not just use Find & Replace in Word or Excel?

Find & Replace handles one literal character at a time. Cleaning a 5,000-row export of punctuation, symbols, and stray emojis would take dozens of passes and still miss characters you did not anticipate, like curly quotes (U+2019) or non-breaking spaces. Category-based removal catches the entire Unicode class in one pass and tells you how many it found.

Is there a size limit, and does my text leave the browser?

Everything runs locally in your browser — nothing is uploaded. Inputs beyond about 1 MB still work; the tool defers recomputation so typing stays responsive, and you will see a notice when you cross that threshold. For multi-megabyte files, paste in sections or download the result rather than copying through the clipboard.

Related tools

Learn more