Skip to content
TabShedKitsMy desk

Furigana Generator – Add Readings to Kanji

Paste Japanese text and get hiragana readings over the kanji, as ruby HTML or plain text. Nothing leaves your browser.

On this device · No upload
Japanese text31 characters
Result
First use downloads the reading dictionary (about 18 MB) once; after that it works offline.

What happens to your text

The text stays in your browser. When you type or paste, a worker thread (so the page keeps responding) cuts the text into words with the kuromoji analyser and the IPAdic dictionary, and asks the dictionary how each word containing kanji is read. The reading is written in hiragana. Where a word has okurigana, the reading is split so only the kanji get the furigana: 行きます becomes 行いきます, not a reading over the whole word.

You get three outputs from the same analysis. Ruby HTML uses the standard <ruby> tags, with <rp> brackets for browsers that cannot show ruby, and a preview under the boxes shows how it will look. Kanji(reading) is plain text such as 今日(きょう), which pastes into chats, notes and spreadsheets. Hiragana only replaces every kanji word with its reading, which is handy for a text-to-speech tool or a young learner.

The first time you use the tool it downloads the dictionary, about 18 MB in twelve files. The service worker keeps it, so the next visit, even offline, starts at once. The dictionary is served from this site, not from a third party, and nothing you type is ever sent anywhere.

Limits to know

  • Up to 300,000 characters at a time. Long text is worked in pieces of about 1,500 characters, so the page stays usable and a progress line shows.
  • A .txt file is read as UTF-8; if it is not, Shift_JIS and EUC-JP are tried. Word and PDF files are not read: copy the text out of them first.
  • The reading is chosen by the dictionary for each word, with no understanding of the story. Words with several readings can be wrong: 生 (なま, せい, い, う), 行った (いった or おこなった), 人気 (にんき or ひとけ), 上手 (じょうず or うわて). Check anything you will print or teach.
  • Names of people and places that are not in the dictionary may get a guess or no reading at all. Words with no reading stay as they are.
  • Particles are left alone: は stays は, and furigana is never added over hiragana or katakana.
  • The ruby preview shows the first few thousand characters only; the copy and download buttons use the full result.

Worked example: a reading sheet for a class

The page opens with two sample sentences already converted. In 今日は東京へ行きます the tool returns 今日きょうは東京とうきょうへ行いきます. Notice that the particles は and へ are untouched and the okurigana き is outside the furigana.

  1. Paste the paragraph you want to teach from, or drop a .txt file onto the left box.
  2. Read the preview. Correct any wrong reading in the Ruby HTML or Kanji(reading) text by hand; the tool cannot know that 生 should be せい in your sentence.
  3. Press Copy on the Ruby HTML tab and paste it into a web page, an email in HTML or a Google Docs HTML import. For a worksheet in Word, copy from the preview box instead, which keeps the ruby layout in most editors.
  4. For a vocabulary list, use the Kanji(reading) tab and paste the lines into a spreadsheet.

Problems and fixes

  • "The dictionary could not be loaded." The first run needs the network for about 18 MB. Check your connection or any blocker that stops requests to this site, then try again.
  • One reading spans two kanji. The reading belongs to the whole word, so 今日 shows きょう over both characters. Many words (今日, 大人, 田舎) cannot be split kanji by kanji, which is why the tool does not try.
  • Ruby does not show in my editor. Not every editor renders ruby. Use the Kanji(reading) tab, which works everywhere.
  • Chinese text in the box. Hanzi that are not Japanese words get no reading. Use Chinese to Pinyin for Chinese.
  • Rōmaji instead of kana? This tool writes hiragana. For Korean text see Korean Romanization.

Credits and licences

Word analysis: @sglkc/kuromoji 1.1.0, a browser-friendly fork of kuromoji.js by Takuya Asano, Apache License 2.0. Dictionary: mecab-ipadic 2.7.0, copyright 2000-2003 Nara Institute of Science and Technology, used under its own licence (use, reproduction and distribution permitted with the notice; the full text is at /vendor/kuromoji/NOTICE.md). See also open data and libraries on the About page.

How to use Furigana Generator

  1. Paste your text

    Type or paste Japanese text, or drop a .txt file onto the left box. A sample is already converted, and the result updates as you type.

  2. Pick the output

    Ruby HTML for web pages, Kanji(reading) for chats and spreadsheets, Hiragana only to replace every kanji with its reading.

  3. Copy or download

    Press Copy or Download. The first use loads an 18 MB reading dictionary once; after that the tool works offline.

Frequently asked questions

Is my text uploaded?
No. The analysis runs in your browser, in a background thread. Only the dictionary is downloaded, from this site.
How accurate are the readings?
Most everyday text is right, but the dictionary picks a reading per word without understanding the story. Words with several readings, names and rare words can be wrong, so check anything you will print or teach.
Does it work offline?
Yes, after the first use. The dictionary (about 18 MB) is kept by your browser through the service worker.
Why is furigana not added over hiragana?
Readings are only needed for kanji. Particles such as は and へ, okurigana and katakana words are left as written.
Is there a size limit?
Up to 300,000 characters at a time. Long text is processed in pieces so the page keeps responding.