Nouplo

Furigana generator

Readings over the kanji of any Japanese text, worked out on your own device.

Examples temporarily replace the input and settings. You can restore your previous work.

0 characters
Words given a reading
0
Words it could not read
0
Output
Paste or type above to see the result here.

Your files stay on your device

This tool runs entirely in your browser. What you paste is processed on your device and is never sent to a server, so it is safe to use with tokens, keys and unpublished text.

Example

Example: furigana for one sentence, written the five ways the tool offers

A learner’s sentence: 明日は友達と東京駅で会います。(“Tomorrow I’m meeting a friend at Tokyo Station.”) Five words get a reading; the kana between them stays as it is.

  1. 会います gets its reading over 会 only, 会(あ)います: the okurigana stays outside.
  2. 東京駅 is read as two words, 東京 and 駅, each with its own reading.
Output Result
Readings in brackets 明日(あした)は友達(ともだち)と東京(とうきょう)駅(えき)で会(あ)います。
HTML ruby tags each word in a ruby element with its reading in rt, such as 明日 over あした; a web page shows the reading above the kanji
|漢字《かんじ》 |明日《あした》は|友達《ともだち》と|東京《とうきょう》|駅《えき》で|会《あ》います。
Everything in hiragana あしたはともだちととうきょうえきであいます。
Everything in katakana アシタハトモダチトトウキョウエキデアイマス。

Words given a reading: 5; words it could not read: 0. The brackets version suits an email or a printout, the ruby version a web page.

How it works

  1. Paste or type Japanese text. The first time, the Japanese dictionary (about 18 MB) loads once; after that the result follows as you type.
  2. Choose how the readings are written: brackets for print and email, ruby tags for a web page, |漢字《かんじ》 for Aozora Bunko and Japanese novel sites.
  3. If a word comes out with the wrong reading, write it as |漢字《よみ》 in the input and that reading is used.
  4. Copy the result.

Frequently asked questions

How accurate are the readings?
They come from a morphological analyser, kuromoji with the IPADIC dictionary. Against 5,000 sentences whose readings were checked by hand (the JSUT corpus), about four sentences in five without numbers came out entirely right, and about 1.7 % of kana were wrong. The mistakes are words with more than one reading — 開いた is あいた or ひらいた depending on what opened — and the names of people and places. Read it through before you rely on it, especially for study.
Can I fix a wrong reading?
Write it in the input as |東京《とうけい》: what stands between the bar and 《 is the text, what is inside 《》 is its reading, and that reading is used as it is. When the text is all kanji the bar can be left out (東京《とうけい》). It is the notation of Aozora Bunko and the Japanese novel sites, so it can stay in a manuscript.
What about numbers and dates?
Numbers stay as digits, and the counter after them is read to match: 9月 is がつ, 3分 is ぷん and 3本 is ぼん. When the reading says the number itself — 1日 is ついたち, 20日 is はつか, 1人 is ひとり — it covers the digits too, so in hiragana 9月1日 becomes 9がつついたち, not 9がつ1ついたち. Numbers written in kanji (三月) are read by the dictionary.
Does the okurigana get furigana too?
No. The reading goes over the kanji only and the kana that follow stay as they are: 引き出し becomes 引(ひ)き出(だ)し, not the whole word with the whole reading on top.
Can I get romaji?
Not here: this writes kana, which is what furigana is. For a name in the romanisation a Japanese passport uses, there is Passport romaji.
Does every kanji get furigana, or only the hard ones?
Every word with kanji gets its reading: there is no school grade to leave the easy ones out. In 明日は友達と東京駅で会います all five get one. Delete any readings you don’t need from the result.
Is my text uploaded?
No. The dictionary is loaded into your browser once and every reading is worked out on your own device. After the first load it also works offline.

Made for a specific use

Related tools