Context-aware Japanese furigana using Sudachi and ModernBERT
Details
- External ID
- 48322203
- Source
- HN
- Company
- —
- Product
- Context-aware Japanese furigana using Sudachi and ModernBERT
- Website domain
- ezfurigana.com
- Launched
- May 29, 2026
- Cohort
- —
- Upvotes
- 37
- Upvotes percentile
- 0.8109854604200323
- Tags
- —
- Fetched at
- Sept. 7, 2026, 9:26 p.m.
- Updated at
- Sept. 7, 2026, 9:26 p.m.
Enrichment
- Theme
- web development and browser utilities
- Vertical
- Education
- Function
- Content generation
- Audience
- B2C
- AI stance
- AI-native
- Project type
- Hobby / open-source project
- Normalized one-liner
- japanese furigana generation using ml
- Manually corrected
- False
Could you build this?
Partial The web interface and document parsing are simple, but accurately disambiguating context-dependent Japanese kanji readings requires training and serving a fine-tuned transformer model.
What it would actually take: The pipeline pairs a morphological dictionary like SudachiPy with a fine-tuned ModernBERT model served via ONNX Runtime or PyTorch for disambiguating ambiguous readings. The hard engineering involves collecting and cleaning large Japanese annotated reading corpora, training the model for sequence classification, and optimizing token inference latency to process uploaded PDFs and documents in real time.
Discussion
20 comments analyzed.
Competitors mentioned: Rikaikun/Rikaichan, tkgje.jp (Japanese-English dictionary)
Concerns raised: Corner cases and accuracy on context-dependent readings, High token consumption from Claude API calls, Personal names and place names still problematic, Rendaku handling incomplete
Feature requests: Japanese language UI/support, Pitch accent support, Improved handling of adjacent token merging
Competitors
Other products that read as similar to this one — 830 launches clear the similarity bar, closest 8 shown.
Attention rank: #158 of 831 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).
Launched 211 days after the earliest competitor.
- Context-Aware AI Assistant for macOS [Open Source] · hn · 2026-01-16 · 6 upvotes · similarity 0.50
- Ilha · hn · 2026-04-16 · 23 upvotes · similarity 0.46
- Zumly · hn · 2026-04-03 · 5 upvotes · similarity 0.46
- AI Anime Finder · hn · 2026-07-25 · 5 upvotes · similarity 0.46
- boar-app · github · 2026-09-22 · 14 upvotes · similarity 0.46
- Generative UIs for the Web (Experimental) · hn · 2026-01-21 · 5 upvotes · similarity 0.46
- Linguin, a live multilingual dictionary for 120 languages · hn · 2026-07-20 · 5 upvotes · similarity 0.45
- window-sweaters · github · 2026-09-15 · 587 upvotes · similarity 0.45
Other launches for this product
- No other launches for this product.
Same idea, different domain
Nobody's really built a content generation tool for Government yet.