How it works
Read the word. See its reading.
Paste Japanese text and choose Add annotations. Kanji words receive a hiragana reading above them using native ruby annotations. The original text, spaces, punctuation, and line breaks are preserved.
Common katakana loanwords receive a suggested source spelling, such as コンピューター → computer, パン → pão (Portuguese), or アルバイト → Arbeit (German). Non-English sources are labeled. This is a small built-in word list, not a complete etymological dictionary or translation service.
Unknown katakana, names, sound effects, and Japanese-made foreign expressions may have no recoverable original. Unmatched words remain unchanged and are underlined for review. Add a custom annotation to supply a spelling or fix a reading.
- Enter text or try the Japanese sample.
- Optionally add corrections, then choose Add annotations.
- Read the result, copy annotated text, or download an HTML file.
Good to know
A reading aid.
Are the readings always correct?
No. Context-dependent readings, names, and unusual spellings can be wrong. Furigana is placed only above kanji; written kana remains unannotated. If a mixed word cannot be aligned unambiguously, its kanji are flagged for review. Use custom annotations for corrections.
Why not restore every foreign word?
Katakana does not uniquely encode the source language or spelling. Source labels are representative suggestions, not proof of a particular borrowing history. Unsupported words are not guessed.
Does it use a paid AI service?
No. The browser downloads an open-source Japanese dictionary and processes text in a background worker. The roughly 18 MB dictionary is loaded only when you run this tool. Your browser may cache it.
What software powers this tool?
Kuromoji.js 0.1.2 and its IPADIC dictionary. See the software license and dictionary notices.