{"record": {"aliases": ["Sarashina2.2-TTS: Tackling Kanji Polyphony in Japanese Speech Generation via Data Scaling and Targeted Data Synthesis"], "artifacts": [{"id": "arxiv:2606.25369", "kind": "paper", "url": "https://arxiv.org/abs/2606.25369"}, {"id": "gh:sbintuitions/sarashina2.2-tts", "kind": "repo", "url": "https://github.com/sbintuitions/sarashina2.2-tts"}], "categories": ["General AI", "Speech & Audio", "cs.SD"], "description": {"en": "A Japanese speech-synthesis pronunciation benchmark covering the officially designated Joyo kanji and their readings."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2606.25369", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2606.25369", "title": "Joyo Kanji Yomi Benchmark"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "key": "claire-radar:2606.25369", "modality": null, "name": "Joyo Kanji Yomi Benchmark", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"admission_policy_version": "2026-09-09.1", "confidence": "Medium", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-09-25T04:40:22Z", "data_status": "primary-source-indexed", "display_eligible": "true", "origin_record_id": "bm_sarashina2-2-tts-tackling-kanji-polyphony-_d9f58013", "origin_source": "arxiv", "recognition_confidence": "0.55", "record_sha256": "6c2ee7c8765a85cc1e5d1113b6fbd60adf12907eeb8b942125a2be69adac12cb", "relation": "evaluates_only", "review_model": "codex-cli-default", "review_state": "ai-reviewed", "reviewed_at": "2026-09-24T20:22:34.615233Z", "source_url": "https://arxiv.org/abs/2606.25369"}, "publisher": null, "released": "2026-06-24", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2606.25369", "source_url": "https://arxiv.org/abs/2606.25369"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2606-25369", "source": "claire_radar", "source_benchmark_id": "2606.25369", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": ["Sarashina2.2-TTS: Tackling Kanji Polyphony in Japanese Speech Generation via Data Scaling and Targeted Data Synthesis"], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Speech & Audio", "attentionForecast": {"confidence": "Medium", "horizon": "7d", "reason": "Comprehensive Joyo-kanji coverage and a pronunciation-specific metric may attract focused interest from Japanese TTS researchers.", "score": 61}, "capabilities": [], "confidence": "Medium", "construction": "Unknown", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-09-24T20:22:34.615233Z", "inputHash": "ef263122325951c7f305eb5a7b3855f10ee85cab645dc42edbc62435f8730641", "model": "codex-cli-default"}, "curation": {"admissionPolicyVersion": "2026-09-09.1", "canonicalNameEvidence": "we introduce the Joyo Kanji Yomi Benchmark", "canonicalNameSource": "abstract", "decisionReason": "The source provides a formal benchmark name, defined coverage, a concrete error metric, and a direct public benchmark link.", "evaluationEvidence": {"reuse": "https://github.com/sbintuitions/JoyoKanji-Yomi-Benchmark", "scoring": "Kana-CER, a metric that compares synthesized speech against reference readings in the kana space", "task": "covering all 2,136 Joyo kanji and their 4,378 readings"}, "model": "codex-cli-default", "reviewedAt": "2026-09-24T20:22:34.615233Z", "state": "ai-reviewed"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "A Japanese speech-synthesis pronunciation benchmark covering the officially designated Joyo kanji and their readings.", "displayEligible": true, "domainCuration": {"method": "rules-v1", "state": "auto"}, "evaluationMode": "public_reusable", "evidence": {"reasonCodes": ["benchmark term in abstract", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL", "named entity is a model, method, framework, or dataset used in evaluation"], "snippet": "Second, we introduce the Joyo Kanji Yomi Benchmark (https://github.com/sbintuitions/JoyoKanji-Yomi-Benchmark), covering all 2,136 Joyo kanji and their 4,378 readings."}, "familyId": "bmf_795c5fd10623", "firstSeenAt": "2026-08-19", "heat": null, "id": "bm_sarashina2-2-tts-tackling-kanji-polyphony-_d9f58013", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/sbintuitions/sarashina2.2-tts", "data": null, "paper": "https://arxiv.org/abs/2606.25369", "pdf": "https://arxiv.org/pdf/2606.25369", "project": null, "report": "https://arxiv.org/abs/2606.25369"}, "metrics": [], "motivation": "While large language model (LLM)-based text-to-speech (TTS) systems have achieved high-quality speech synthesis, most existing systems focus on English and Chinese.", "name": "Joyo Kanji Yomi Benchmark", "oneLine": "A Japanese speech-synthesis pronunciation benchmark covering the officially designated Joyo kanji and their readings.", "paperTitle": "Sarashina2.2-TTS: Tackling Kanji Polyphony in Japanese Speech Generation via Data Scaling and Targeted Data Synthesis", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2606.25369", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-08-19T11:05:13.395391Z"}, "publications": [], "readiness": "Runnable", "recognitionConfidence": 0.55, "relation": "evaluates_only", "releasedAt": "2026-06-24", "reviewLifecycle": {"state": "pending"}, "source": {"authors": ["Lianbo Liu", "Shiao Zhu", "Kai Washizaki", "Reo Yoneyama", "Haesung Jeon", "Mengjie Zhao", "Yusuke Fujita", "Hao Shi", "Nao Yoshida", "Yuan Gao", "Roman Koshkin", "Yukiya Hono", "Yui Sudo"], "categories": ["cs.SD", "cs.AI", "cs.CL", "eess.AS"], "id": "2606.25369", "title": "Sarashina2.2-TTS: Tackling Kanji Polyphony in Japanese Speech Generation via Data Scaling and Targeted Data Synthesis", "type": "arxiv", "url": "https://arxiv.org/abs/2606.25369"}, "sourceUpdatedAt": "2026-06-24T03:57:21Z", "topics": ["cs.SD"], "venueAttempts": [], "whyItMatters": "It targets context-dependent kanji pronunciation and uses kana-space character error rate to reduce scoring distortions from orthographic variation."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
