Normalize text to NFC, NFD, NFKC or NFKD, see what changed, or strip accents
Runs in your browser. Nothing you type is sent anywhere. Checked against sources on 16 Nov 2026.Checked against the tool’s code on 18 November 2026. Figures read from their sources on 16 November 2026. What we collect
A Unicode normalizer rewrites text in one standard form so equal-looking strings compare equal. In the example, NFC joins the split "e" and its accent into one character, 15 code points to 14; NFKC also turns fullwidth letters, the "fi" ligature and a circled 1 into plain text, giving "Café Tea file 1" and 26 bytes down to 16.
| Form | What it does | Code points | UTF-8 bytes |
|---|---|---|---|
| NFC | joins a letter and its accent into one character where Unicode has one | 15 to 14 | 26 to 25 |
| NFD | splits each accented letter into the letter and its combining accent | 15 to 15 | 26 to 26 |
| NFKC | also turns fullwidth, circled, superscript and ligature forms into plain characters, then composes | 15 to 15 | 26 to 16 |
| NFKD | turns compatibility forms into plain characters and splits accents off | 15 to 16 | 26 to 17 |
Nothing is uploaded: this runs in your browser.
Paste the text into the box; an example with a split accent and fullwidth letters is already there.
Pick the form: NFC for storage and comparison, NFKC to fold fullwidth and styled forms too.
Read the table of changes: each character before and after, with its code points.
Press Copy result, or paste a second string under Compare with to see why the two do not match.
Unicode often has two ways to write the same thing. The letter é can be one character, U+00E9, or two: a plain e followed by the combining acute accent U+0301. They look identical and print identically, but a database lookup, a file name check or a password compare sees different bytes and says they differ. Text pasted from a Mac, a PDF or a phone keyboard is a common source of the split form.
Normalization picks one spelling. NFC composes (one character where Unicode has one) and is what the web and most databases expect. NFD decomposes. NFKC and NFKD also fold compatibility forms, such as fullwidth T, the fi ligature, superscript ² and circled ①, into plain characters, which is right for search and user names but loses formatting.
| Question | Answer |
|---|---|
| Forms | NFC, NFD, NFKC and NFKD, as defined by Unicode Standard Annex #15 |
| What it shows | Each changed character before and after, with code points, and the byte counts |
| Remove accents | Decomposes, drops the combining marks: "Crème brûlée à São Paulo" becomes "Creme brulee a Sao Paulo" |
| Compare | Says whether two strings are equal as typed, after NFC, or only after NFKC |
| Where it runs | In your browser, with its own Unicode tables; nothing is uploaded |
Checked on 16 November 2026 against Unicode Standard Annex #15. Page last updated on 16 November 2026.
Store and compare in NFC unless you have a reason not to. Use NFKC for search keys and identifiers, where a fullwidth or styled letter should match its plain twin. For lookalike letters from other scripts, which no normalization form changes, use the homoglyph detector; for hidden characters, the Invisible Character Detector.
A Unicode normalizer rewrites text in one of the four standard forms, so the same text always has the same bytes. NFC joins a letter and its accent into one character; NFKC also turns fullwidth, ligature and circled forms into plain letters. Strings that looked equal then compare equal in code and databases.
Both keep the meaning and the look. NFC composes: é is stored as the single character U+00E9. NFD decomposes: é becomes e followed by the combining accent U+0301. The example shows NFC turning 15 code points into 14. Most web pages, databases and Windows expect NFC.
NFKC does everything NFC does and also replaces compatibility characters with their plain equivalents: fullwidth T becomes T, the fi ligature becomes fi, ① becomes 1 and ² becomes 2. The example becomes "Café Tea file 1". Use NFKC for search and usernames, NFC for storing text as written.
One of them probably holds a split accent or a compatibility character. Paste the first string in the box and the second under Compare with: the page says whether they are equal as typed, equal after NFC, or equal only after NFKC. Normalize both sides before comparing to fix the bug.
Tick Remove accents. The text is decomposed and every combining mark is dropped, so "Crème brûlée à São Paulo" becomes "Creme brulee a Sao Paulo". Letters with no separate accent, such as ø or ß, are kept as they are, since Unicode does not split them.
No. A Cyrillic а (U+0430) is a different letter from the Latin a in every normalization form, so NFKC keeps it. Lookalike letters are found by a homoglyph detector instead, which compares characters against the Unicode confusables list and names the letter each one imitates.
JavaScript has text.normalize('NFC'); Python has unicodedata.normalize('NFC', text); Java has java.text.Normalizer. This page uses the same JavaScript call, so the result here matches what your browser code produces. The table of changes helps when you need to see which characters a call altered.
Rates and rules change; check the source before you rely on a figure.
Still stuck? Ask about Unicode Normalizer: NFC, NFD, NFKC and NFKD Online and anybody who knows can answer. No account needed.
Same job, other tools: Remove Tracking Parameters – Strip utm, fbclid and gclid from a link before you share it · JSON Minifier – Strip whitespace from JSON for production · Base64 Decode – Turn Base64 back into text or a downloadable file · Base64 Encode – Turn text or a file into Base64.
On the related tools: JWT Decoder – Paste a token to read its header, payload and expiry.
On the same topic: Invisible Character Detector – Find zero-width spaces, odd spaces, bidi controls and lookalike letters, then clean them · Homoglyph Detector – Homoglyph detector: Cyrillic, Greek and styled letters posing as Latin in names and links · Invisible Character Copy and Paste: Blank Text – Copy a blank character in one tap: Hangul filler, braille blank, zero-width space.