Kana to Romaji
Convert hiragana and katakana to Hepburn romaji, handling small tsu, ん, and long vowels.
Small っ doubles the next consonant, ん takes an apostrophe before a vowel or y, and ー lengthens the vowel before it.
Kanji is not converted, because reading it requires knowing the word. Long vowels written as おう or うう are left as written, since only a dictionary can tell とうきょう (Tōkyō) from a word where the two vowels genuinely belong to different syllables.
Hepburn romanisation maps each kana to the Latin letters an English speaker would read closest to the Japanese sound. がっこう becomes gakkou because the small っ doubles the following consonant, and トーキョー becomes tōkyō because ー lengthens the vowel before it.
Why Hepburn rather than the alternatives
Japanese has three romanisation systems in circulation. Hepburn spells sounds the way an English speaker would read them, which is why し is shi and つ is tsu. Kunrei-shiki, the system taught in Japanese primary schools, keeps the kana table regular instead, giving si and tu. Nihon-shiki is older and more regular still.
Hepburn is what passports, road signs, and almost all English-language publishing use, so it is the default here. If you have seen a name spelled two different ways — Itiro against Ichiro — the difference is usually Kunrei against Hepburn rather than an error.
The three rules that trip people up
Most kana map one to one, and the interesting behaviour is at the edges. The small っ is not a sound of its own: it holds the following consonant, so きって is kitte. Before ch it is written t rather than c, giving matchi for マッチ.
ん is a syllable in its own right, and before a vowel or y it needs an apostrophe to stay readable — じゅんあい is jun’ai, not junai, which would be read as a different word. And the katakana ー lengthens the preceding vowel, written either with a macron (ō) or as a repeated vowel (oo) depending on house style.
Worked example: がっこう
Read the kana in order, applying the gemination rule in the middle:
- 1 Take the kana one at a time. が, っ, こ, う — four characters, but only three sounds.
- 2 Romanise the first. が is ga.
- 3 Hold the consonant at the small っ. The small tsu doubles whatever consonant follows it, so the k of こ is written twice.
- 4 Continue through the rest. こ is ko and う is u, giving gakkou.
- 5 Decide how to write the long vowel. Some styles write this word gakkō with a macron. This converter leaves おう as written, because only a dictionary knows when the two vowels belong together.
The rules this converter applies
Everything else is a direct one-to-one or two-to-one mapping from the kana table.
| Input | Rule | Example |
|---|---|---|
| っ or ッ | Doubles the next consonant | きって → kitte |
| っ before ch | Written t, not c | マッチ → matchi |
| ん before a vowel or y | Takes an apostrophe | じゅんあい → jun’ai |
| ー (katakana) | Lengthens the vowel before it | ラーメン → rāmen |
| きゃ, しゃ, ちゃ … | Two kana, one syllable | きょうと → kyouto |
| ファ, チェ, ヴィ … | Katakana sounds for foreign words | ファイル → fairu |
What a converter cannot decide for you
Two things genuinely require knowing the word rather than the characters. Kanji is left untouched, because the same character has different readings in different words — 生 alone has more than ten. Any tool that appears to romanise kanji is either using a dictionary or guessing.
The second is おう and うう. In とうきょう the おう is a long o, conventionally written Tōkyō. But in words where the two vowels belong to different parts — such as 子牛, koushi — writing them as a long vowel would be wrong. Since the characters look identical, this converter leaves them as written rather than guessing.
There is also a small variation worth knowing: traditional Hepburn writes ん as m before b, m, and p, giving sempai where modified Hepburn gives senpai. Modified Hepburn, used here, is the more common modern convention, but both appear in print and neither is wrong.