The new furigana converter stitches Japanese text with phonetic guides by leveraging Sudachi’s tokenization and ModernBERT’s contextual embeddings. Users upload plain text, PDFs, images, EPUBs, subtitles, or URLs and instantly generate hiragana, katakana, or romaji annotations. The tool runs entirely in the browser, keeping uploads private and deleting them after 24 hours for students and language learners daily practice sessions.

A standout feature is JLPT filtering, which suppresses furigana on lower‑level kanji while highlighting N3 and above. This lets beginners focus on common characters and gradually tackle harder readings. The interface also saves clicked words locally, allowing learners to build flashcards or review vocabulary without a backend account for daily study sessions and track progress over time within the browser.

Because the entire pipeline runs client‑side, the service remains free and ad‑free. Developers can embed the same Sudachi‑BERT logic into their own applications, opening avenues for AI‑enhanced reading tools across education and publishing. The converter already supports multiple output formats, from plain text to Anki decks, making it a practical resource for anyone teaching or learning Japanese today with no.