Language Support¶

translit ships with 83 built-in language profiles that provide language-specific transliteration rules. You can also register custom profiles at runtime.

Built-in languages¶

European languages¶

Southeast Asian languages¶

Semitic languages¶

Iranian languages¶

Other Middle Eastern languages¶

Ethiopic languages¶

African languages¶

Caucasian languages¶

Indic languages¶

All 10 Brahmic scripts use virama/mātrā-aware transliteration: consonants carry an inherent "a" that is suppressed by virama (halant) or replaced by dependent vowel marks.

Tibetan languages¶

Southeast Asian languages¶

Philippine languages¶

Americas¶

Lisu¶

East Asian & other non-European languages¶

Toned pinyin: Pass tones=True to transliterate() for diacritical pinyin output (e.g., "běi jīng" instead of "bei jing"). Coverage includes the ~2,000 most common characters.

CJK examples¶

from translit import transliterate, slugify

# Chinese
transliterate("北京市")             # => "bei jing shi"
slugify("北京烤鸭")                # => "bei-jing-kao-ya"

# Korean
transliterate("서울")               # => "seo ul"
slugify("대한민국")                # => "dae-han-min-gug"

# Japanese (hiragana/katakana use Hepburn; kanji use Chinese pinyin)
transliterate("ひらがな")           # => "hiragana"
transliterate("東京タワー")         # => "dong jing tawa-"
transliterate("東京タワー", lang="ja")  # => "dong jing tawa" (ー dropped)

Reverse transliteration¶

translit can convert romanized Latin text back to native script for selected languages using the target parameter:

from translit import transliterate, reverse_langs

transliterate("Moskva", target="ru")    # → "Москва"
transliterate("Kyiv", target="uk")      # → "Київ" (approximate)
transliterate("Athina", target="el")    # → "Αθηνα"

# List supported languages
reverse_langs()                         # → ["el", "ru", "uk"]

Reverse transliteration uses greedy longest-match scanning to handle digraphs and trigraphs (e.g., "shch" → щ). See Limitations for round-trip degradation details.

Auto-detecting language from script¶

When you don't know the language of the input text, pass lang="auto" to automatically detect the dominant non-Latin script and select the appropriate language profile:

from translit import transliterate, slugify, LANG_AUTO

# Detects Cyrillic → uses Russian ("ru") profile
transliterate("Москва", lang="auto")         # => "Moskva"

# Detects Thai → uses Thai ("th") profile
transliterate("ภาษาไทย", lang="auto")         # => Thai transliteration

# Detects Devanagari → uses Hindi ("hi") profile
transliterate("नमस्ते", lang="auto")           # => "namaste"

# Detects Hangul → uses Korean ("ko") profile
slugify("한국어", lang="auto")                 # => Korean romanization slug

# Works with all call sites
from translit import TextPipeline, Slugifier

pipe = TextPipeline(transliterate=True, lang="auto")
pipe("こんにちは")    # => Japanese transliteration

s = Slugifier(lang="auto")
s("東京タワー")      # => CJK slug

How auto-detection works¶

1. Scans the input for the first non-Latin, non-Common character
2. For ambiguous scripts, scans for exclusive discriminator characters
3. Maps the detected script (and discriminated language) to a language code
4. Falls back to default (no language override) if the text is Latin-only or the script has no mapping

For a detailed walkthrough of the three-stage detection pipeline, discriminator tables, and fail-safe guarantees, see Language Detection.

Script-to-language mapping¶

For unambiguous scripts (one script = one language), detection is immediate:

Character-level discrimination for ambiguous scripts¶

For scripts shared by multiple languages, translit scans for exclusive characters — codepoints that appear in exactly one language's alphabet among the profiles we support:

If no exclusive characters are found, the script default is used (Cyrillic → ru, Arabic → ar, Latin → no override). If exclusive characters from two different languages appear in the same text (e.g., Ukrainian ї and Serbian ћ), detection falls back to the script default — this is the fail-safe guarantee.

# Ukrainian detected by exclusive ї
transliterate("Київ", lang="auto")   # → uses uk profile

# Persian detected by exclusive پ
transliterate("پارسی", lang="auto")  # → uses fa profile

# German detected by ß
transliterate("Straße", lang="auto") # → uses de profile

# No exclusive chars → safe default
transliterate("Москва", lang="auto") # → uses ru (unchanged)

For scripts that remain ambiguous after discrimination (Devanagari, Han), pass an explicit language code when accuracy matters.

from translit import LANG_AUTO, transliterate
transliterate("Москва", lang=LANG_AUTO)

Using language profiles¶

With functions¶

from translit import transliterate, slugify, sanitize_filename

transliterate("Ürümqi", lang="de")         # => "Ueruemqi"
slugify("Ärger im Büro", lang="de")        # => "aerger-im-buero"
sanitize_filename("Ärger.txt", lang="de")  # => "Aerger.txt"

With classes¶

from translit import Slugifier, TextPipeline

slug = Slugifier(lang="de", separator="_")
pipe = TextPipeline(transliterate=True, lang="fr")

Language constants¶

Pre-defined constants for type safety:

from translit import LANG_DE, LANG_FR, transliterate

transliterate("Ä", lang=LANG_DE)  # => "Ae"
transliterate("Ç", lang=LANG_FR)  # => "C"

Listing available languages¶

from translit import list_langs

print(list_langs())
# => ['am', 'ar', 'as', 'bg', 'bn', 'bo', 'ca', 'cs', 'cy', 'da', 'de', 'dv', 'el',
#     'es', 'et', 'fa', 'fi', 'fr', 'ga', 'gu', 'he', 'hi', 'hr', 'hu', 'hy',
#     'is', 'it', 'ja', 'jv', 'ka', 'km', 'kn', 'ko', 'lo', 'lt', 'lv', 'ml', 'mn',
#     'mr', 'mt', 'my', 'ne', 'nl', 'no', 'or', 'pa', 'pl', 'pt', 'ro', 'ru', 'sa',
#     'si', 'sk', 'sl', 'sq', 'sr', 'sv', 'ta', 'te', 'th', 'tr', 'uk', 'vi', 'zh']

Custom language profiles¶

register_lang¶

Register a new language profile or override an existing one:

from translit import register_lang, transliterate

# Register Esperanto
register_lang("eo", {
    "ĉ": "cx",
    "ĝ": "gx",
    "ĥ": "hx",
    "ĵ": "jx",
    "ŝ": "sx",
    "ŭ": "ux",
})

transliterate("ĉapelo", lang="eo")  # => "cxapelo"

register_replacements¶

Register global pre-transliteration string replacements:

from translit import register_replacements, transliterate

register_replacements({
    "©": "(c)",
    "®": "(R)",
    "™": "(TM)",
})

transliterate("Hello™ World©")  # => "Hello(TM) World(c)"

Norwegian variants¶

Both "no" and "nb" (Bokmål) map to the same Norwegian profile. "nn" (Nynorsk) also uses the same mappings. Use any of these codes interchangeably.

Historical and ancient scripts¶

translit includes transliteration mappings for several historical and ancient writing systems:

These mappings provide approximate romanizations suitable for search indexing and display purposes. They are not intended as scholarly transliteration standards.