Remove Accents

Unicode can write é as one character or as e followed by a combining acute accent. The tool first applies NFD normalization, which splits every accented letter into its base letter plus combining marks, then deletes the marks (the Combining Diacritical Marks blocks, U+0300–U+036F and related ranges). This handles French, Spanish, Portuguese, Vietnamese, Czech, Polish, Romanian, Turkish and most other Latin-script languages in one step.

Marks are only removed from Latin and Greek letters (and Cyrillic when you transliterate), so Hindi, Thai, Arabic and other scripts that need their vowel signs are never damaged.

Some letters look accented but are separate letters in Unicode, so normalization leaves them alone. With "Special letters" on (the default), the tool maps them exactly as the URL slug generator does:

German readers expect ä, ö and ü as ae, oe and ue, so Müller becomes Mueller; turn on "German umlauts" for German names. For other languages leave it off. Our transliteration guide covers language conventions in detail.

• Keep (default): Chinese, Cyrillic, Greek and other scripts stay as they are. Greek still loses its tonos accents, so Καλημέρα becomes Καλημερα.
• Transliterate: Cyrillic and Greek are romanized with the slug generator's tables, so Хрущёв becomes Khrushchyov. Other scripts are kept.
• Remove: every non-Latin letter is deleted and leftover spaces are collapsed, leaving only the Latin text.

Both give the tool's default result (checked against it for every test string):

// Same table as the tool: letters that Unicode does not split into base + accent.
const SPECIAL = { ß: 'ss', ẞ: 'SS', æ: 'ae', Æ: 'AE', œ: 'oe', Œ: 'OE', ø: 'o', Ø: 'O', đ: 'd', Đ: 'D', ð: 'd', Ð: 'D', þ: 'th', Þ: 'TH', ł: 'l', Ł: 'L', ı: 'i', ħ: 'h', Ħ: 'H', ŧ: 't', Ŧ: 'T', ŀ: 'l', Ŀ: 'L', ŋ: 'n', Ŋ: 'N', ſ: 's', ĸ: 'k', ƒ: 'f' };
const mapSpecial = (s) => [...s].map((c) => SPECIAL[c] ?? c).join('');

function removeAccents(text) {
  const s = mapSpecial(text.normalize('NFC')).normalize('NFD')
    // Drop combining accents only after a Latin or Greek letter, so other scripts stay intact.
    .replace(/([\p{Script=Latin}\p{Script=Greek}])[\u0300-\u036f\u1ab0-\u1aff\u1dc0-\u1dff\u20d0-\u20ff\ufe20-\ufe2f]+/gu, '$1');
  return mapSpecial(s).normalize('NFC');
}

import unicodedata

# Same table as the tool: letters that Unicode does not split into base + accent.
SPECIAL = {'ß': 'ss', 'ẞ': 'SS', 'æ': 'ae', 'Æ': 'AE', 'œ': 'oe', 'Œ': 'OE', 'ø': 'o', 'Ø': 'O', 'đ': 'd', 'Đ': 'D', 'ð': 'd', 'Ð': 'D', 'þ': 'th', 'Þ': 'TH', 'ł': 'l', 'Ł': 'L', 'ı': 'i', 'ħ': 'h', 'Ħ': 'H', 'ŧ': 't', 'Ŧ': 'T', 'ŀ': 'l', 'Ŀ': 'L', 'ŋ': 'n', 'Ŋ': 'N', 'ſ': 's', 'ĸ': 'k', 'ƒ': 'f'}
MARKS = ((0x0300, 0x036F), (0x1AB0, 0x1AFF), (0x1DC0, 0x1DFF), (0x20D0, 0x20FF), (0xFE20, 0xFE2F))


def _is_mark(c):
    return any(a <= ord(c) <= b for a, b in MARKS)


def remove_accents(text):
    s = ''.join(SPECIAL.get(c, c) for c in unicodedata.normalize('NFC', text))
    out, after_latin = [], False
    for c in unicodedata.normalize('NFD', s):
        if _is_mark(c):
            if after_latin:
                continue  # an accent on a Latin or Greek letter
        else:
            name = unicodedata.name(c, '')
            after_latin = 'LATIN' in name or 'GREEK' in name
        out.append(c)
    return unicodedata.normalize('NFC', ''.join(SPECIAL.get(c, c) for c in out))

Need a URL slug rather than clean text? The URL slug generator removes accents, lowercases and joins words with hyphens in one go. If a link already shows %C3%A9-style escapes, decode it with the URL encoder/decoder.

• Unicode Standard Annex #15: Unicode Normalization Forms
• MDN: String.prototype.normalize()
• Python documentation: unicodedata.normalize()