Free Japanese to Romaji Converter

Worked example

Under "Try:" press "Kanji". The box now holds three lines: 私は学生です。, 東京へ行きます。 and 今日はいい天気ですね。 Because the text has kanji, a card titled "Kanji needs a one-time dictionary download" appears with the button "Download dictionary (17 MB)". Until you press it, the "Result" card says "Kanji stay as they are (underlined) until the dictionary above is downloaded, so this result is only partly converted." After the download the result is watashi wa gakusei desu. / toukyou e ikimasu. / kyou wa ii tenki desu ne., with は and へ written as wa and e. Tick "Romaji under each line" to see every Japanese line with its romaji below it. Under "Long vowels", "Macrons (tōkyō)" turns line 2 into tōkyō e ikimasu. but line 3 into kyō wa ī tenki desu ne., so read the second tip before using it.

Which settings to use, and the limits

• Pick the romanization system the document asks for. "Romanization" offers "Hepburn", "Kunrei-shiki" and "Nihon-shiki". Hepburn is the default. For しんぶん ちじみ つづき しゃしん を, Hepburn gives shinbun chijimi tsuzuki shashin o, Kunrei-shiki gives sinbun tizimi tuzuki syasin o, and Nihon-shiki gives sinbun tizimi tuduki syasin wo. Nihon-shiki keeps ぢ and づ apart from じ and ず and writes を as wo. The "Long vowels" menu is shown for Hepburn only; the other two systems always write long vowels as typed.
• Choose a long-vowel style, then read the result. "As typed (toukyou)" is the default. "Macrons (tōkyō)" gives とうきょう おおさか as tōkyō ōsaka. "Dropped (tokyo)" removes the marks. "Passport (ohno, sato)" writes おお as oh and shortens the other long vowels, so おおの さとう ゆうき gives ohno sato yuki. These styles act on the romaji letters, not on the Japanese, so they also shorten pairs that are not long vowels: いい becomes ī and おもう becomes omō with macrons, and ii becomes i when dropped. Use "As typed" for text with such words, or fix them by hand.
• Let the dictionary read kanji, once. Kanji need a dictionary because their readings depend on the word. The download is 17 MB, comes from the site's own server, and is stored in your browser, so on later visits it loads without asking. Private windows may not keep it. Kana-only text never uses it. On a phone, Wi-Fi is the kinder choice. The readings are the dictionary's best guesses: 日本 gives nippon but 日本橋 gives nihonbashi, and 大野さん gives oono-san with a hyphen before the honorific. Check names, place names and anything you will print.
• Keep "Particles as spoken (wa, e, o)" ticked for sentences. It is on by default. It writes は, へ and を as wa, e and o when they are particles, and こんにちは as konnichiwa. With kanji in the text the dictionary decides. Without kanji there is no dictionary, so the tool works from patterns: common words followed by は (わたしは, これは), katakana words followed by は, a lone は or へ with spaces or punctuation around it, and へ right before a motion verb. Untick it for text where は should stay ha. With Nihon-shiki, を stays wo.
• Type romaji to get kana. "To Hiragana" and "To Katakana" take romaji. A short vowel stays short: "tokyo" gives ときょ, while "toukyou" or "tōkyō" gives とうきょう. A macron ō becomes おう, and the other macron vowels get the long vowel mark, so "yūko" gives ゆーこ. An apostrophe or hyphen after n keeps the ん: "shin'ichi" gives しんいち but "shinichi" gives しにち. In "To Hiragana", konnichiwa becomes こんにちは, but "To Katakana" gives コンニチワ. An English word comes out as a mix of kana and Latin letters.
• Use the katakana name helper for forms and cards. Under "Your name in katakana", fill in "First name" and "Last name". "Japanese name (romaji)" reads them as romaji: Hiroshi and Tanaka give ヒロシ タナカ, with "Romaji" Hiroshi Tanaka. Long vowels there are written with macrons, so Yuuko Satou shows as Yūko Satō. "Western name (approximate)" matches sounds by simple rules and joins the parts with ・, giving ジョン・スミス for John Smith. It can be wrong: Michael gives ミチャエル. The name helper is never put into share links.

Common problems and how to fix them

It says "The dictionary could not be downloaded. Check your connection and try again."

The 17 MB kanji dictionary is fetched from the site's own server as 12 files, and each one is checked against its expected size and checksum before it is stored. A dropped connection, a blocked request, a download that sends no data for 20 seconds, a file that fails the size or checksum test, or a browser that cannot unpack gzip files all end here. Press "Try again" to start the download again, or "Cancel" to stop it. Kana-only text keeps working without the dictionary.

Kanji stay in the result, underlined. The tooltip says "This kanji was not converted — it needs the dictionary, or the dictionary does not know the word."

If the card still shows "Download dictionary (17 MB)", the dictionary has not been loaded yet: press it. If the dictionary is loaded and a kanji is still underlined, that word is not in it, which can happen with rare words and some names. Replace only that word with its kana reading and the rest of the line is still read by the dictionary.

A kanji word has the wrong reading, such as 行って giving okonatte or 一人 giving ichinin.

The dictionary chooses one reading for each word from the words around it, and it cannot know which reading you mean. The page warns that names, rare words and lyrics with unusual readings can be wrong. Type that word in kana: 行って can be いって, 一人 can be ひとり and 日本 can be にほん, which gives nihon instead of nippon. Mixed kanji and kana is fine.

A kana sentence leaves へ as "he", for example "gakkouhe ikimasu."

Without kanji the tool has no dictionary. It changes へ to e only when it stands alone between spaces or punctuation, or sits directly in front of a motion verb such as いく, いき or くる. In "がっこうへ いきます" the space breaks the second rule, so the へ stays "he". Type え in place of へ, or untick "Particles as spoken (wa, e, o)" if you want ha, he and wo as written.

"Kana as you type" turned konnichiwa into こんいちわ.

This mode behaves like a Japanese keyboard. Two n's in a row make ん, so "konnichiwa" is read as kon + ichiwa. Type three n's, "konnnichiwa", or use an apostrophe, "kon'nichiwa", to get こんにちわ. A final n stays as a plain "n" until you type the next letter, so "nihon" shows にほn and "nihonn" shows にほん. Lowercase gives hiragana and UPPERCASE gives katakana. Typing wa gives わ, not は.

How it compares

A Japanese keyboard and a dictionary app are better for looking up meanings, picking the right kanji for a name, or checking an unusual reading. This tool converts text you already have: kanji, hiragana and katakana to romaji in three systems, romaji to kana, and a few helpers for forms. It does not translate, add furigana or mark pitch accent. Kanji readings come from a dictionary-based analyser and are best guesses, so for a passport, visa or legal document use the spelling already printed on your official papers. Your text stays in the browser; only the dictionary files are downloaded.