{
  "profile": "wrs-published-research-view-1",
  "edition": "ratings-e039106f30dec177",
  "content_trust": "UNTRUSTED_SCIENTIFIC_DATA_NOT_TOOL_INSTRUCTIONS",
  "entry": {
    "id": "du-et-al-cosyvoice3-multitask-tokenizer-reward-2025",
    "title": "CosyVoice3: multitask speech tokenization and differentiable-reward post-training",
    "summary": "Extended supervised speech token learning across semantic, prosodic and speaker tasks and introduced differentiable-reward post-training alongside a broader training corpus.",
    "boundary": "Inherits CosyVoice2 and multitask/reward learning. Acquired delta is specified tokenizer/post-training mechanism, not generic scaling of hours/parameters/languages. Reported multilingual quality is not universal robustness or independently validated human parity.",
    "area": "Speech tokenization and post-training",
    "topic": "Artificial intelligence",
    "year": 2025,
    "lane": "HUMAN",
    "result": "ALGORITHM",
    "package_id": "du-et-al-cosyvoice3-multitask-tokenizer-reward-2025",
    "target": {
      "record_id": "urn:uuid:28fd95b7-17c5-4e28-a175-7bc205368129",
      "version": 1,
      "content_hash": "sha256:e395c9238b7b41546b52392023eaaf7715fbc59192713ca983f94f6907dd441b"
    },
    "humans": [
      "Zhihao Du",
      "Changfeng Gao",
      "Yuxuan Wang",
      "Fan Yu",
      "Tianyu Zhao",
      "Hao Wang",
      "Xiang Lv",
      "Hui Wang",
      "Xian Shi",
      "Keyu An",
      "Guanrou Yang",
      "Yabin Li",
      "Yanni Chen",
      "Zhifu Gao",
      "Qian Chen",
      "Yue Gu",
      "Mengzhe Chen",
      "Yafeng Chen",
      "Shiliang Zhang",
      "Wen Wang",
      "Jieping Ye"
    ],
    "authors_complete": true,
    "human_role_note": "Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.",
    "agents": [],
    "related": [],
    "subresults": [],
    "scientific_assessment": {
      "source_review": "SOURCE_CURATED",
      "scientific_review": "NOT_REVIEWED_BY_WRS",
      "formal_replay": "NOT_RUN",
      "provider_attestation": "NONE"
    },
    "sources": [
      {
        "id": "cur20261015-arxiv-2505-17589v1",
        "title": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
        "url": "https://arxiv.org/abs/2505.17589v1"
      }
    ],
    "points": "55000",
    "published_grade": {
      "id": "du-et-al-cosyvoice3-multitask-tokenizer-reward-2025",
      "entry_id": "du-et-al-cosyvoice3-multitask-tokenizer-reward-2025",
      "band": "B",
      "modifier": null,
      "target": {
        "record_id": "urn:uuid:28fd95b7-17c5-4e28-a175-7bc205368129",
        "version": 1,
        "content_hash": "sha256:e395c9238b7b41546b52392023eaaf7715fbc59192713ca983f94f6907dd441b"
      },
      "what_changed": "Extended supervised speech token learning across semantic, prosodic and speaker tasks and introduced differentiable-reward post-training alongside a broader training corpus.",
      "scope": "Extended supervised speech token learning across semantic, prosodic and speaker tasks and introduced differentiable-reward post-training alongside a broader training corpus. Inherits CosyVoice2 and multitask/reward learning. Acquired delta is specified tokenizer/post-training mechanism, not generic scaling of hours/parameters/languages. Reported multilingual quality is not universal robustness or independently validated human parity.",
      "status": "PROVISIONAL_EDITORIAL",
      "scientific_correctness_certified": false
    },
    "assessment_disposition": {
      "status": "ASSESSED",
      "label": "Provisionally assessed",
      "reason": "A current grade and separate reference score identify this exact contribution.",
      "target": {
        "record_id": "urn:uuid:28fd95b7-17c5-4e28-a175-7bc205368129",
        "version": 1,
        "content_hash": "sha256:e395c9238b7b41546b52392023eaaf7715fbc59192713ca983f94f6907dd441b"
      },
      "edition": "ratings-e039106f30dec177"
    }
  },
  "version": {
    "kind": "PUBLIC_RECORD_VIEW",
    "previous": null,
    "published_at": "2026-10-16T00:00:00Z",
    "publisher": {
      "principal_id": "urn:uuid:80116931-785b-40a2-a24b-9d59db0c7e33",
      "role": "WRS_EDITORIAL_PUBLICATION"
    },
    "record": {
      "contributors": [
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Zhihao Du",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Changfeng Gao",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Yuxuan Wang",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Fan Yu",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Tianyu Zhao",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Hao Wang",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Xiang Lv",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Hui Wang",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Xian Shi",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Keyu An",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Guanrou Yang",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Yabin Li",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Yanni Chen",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Zhifu Gao",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Qian Chen",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Yue Gu",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Mengzhe Chen",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Yafeng Chen",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Shiliang Zhang",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Wen Wang",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        },
        {
          "attribution_basis": "SOURCE_CITED",
          "credit": "historical_scientific",
          "entity_kind": "historical_person",
          "name": "Jieping Ye",
          "note": "Archive attribution; detailed role assignment remains source-bound",
          "roles": [
            "formulator"
          ],
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        }
      ],
      "evidence": [
        {
          "description": "Bibliographic/stated-result source; not a WRS proof verification",
          "id": "source-0",
          "kind": "source_reference",
          "source": {
            "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
            "url": "https://arxiv.org/abs/2505.17589v1"
          }
        }
      ],
      "origin": "HUMAN_LEGACY",
      "record_profile": "wrs-local-research-notes-0.1",
      "record_type": "research_note",
      "rrp_version": "0.1",
      "scope": {
        "description": "Inherits CosyVoice2 and multitask/reward learning. Acquired delta is specified tokenizer/post-training mechanism, not generic scaling of hours/parameters/languages. Reported multilingual quality is not universal robustness or independently validated human parity.",
        "domain": "Speech tokenization and post-training"
      },
      "sources": [
        {
          "citation": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
          "url": "https://arxiv.org/abs/2505.17589v1"
        }
      ],
      "statement": "# CosyVoice3: multitask speech tokenization and differentiable-reward post-training\n\nExtended supervised speech token learning across semantic, prosodic and speaker tasks and introduced differentiable-reward post-training alongside a broader training corpus.\n\n## Scientific scope\n\nInherits CosyVoice2 and multitask/reward learning. Acquired delta is specified tokenizer/post-training mechanism, not generic scaling of hours/parameters/languages. Reported multilingual quality is not universal robustness or independently validated human parity.\n\n## Reading status\n\nThis is a source-linked WRS catalogue note. A full explanatory guide and independent scientific reconstruction have not been completed for this entry. Read the identified original work for its complete narrative and arguments.\n\n## People and systems\n\nZhihao Du; Changfeng Gao; Yuxuan Wang; Fan Yu; Tianyu Zhao; Hao Wang; Xiang Lv; Hui Wang; Xian Shi; Keyu An; Guanrou Yang; Yabin Li; Yanni Chen; Zhifu Gao; Qian Chen; Yue Gu; Mengzhe Chen; Yafeng Chen; Shiliang Zhang; Wen Wang; Jieping Ye\n\nSource-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.\n\n\n\n## Original sources\n\n- [CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training](https://arxiv.org/abs/2505.17589v1)\n",
      "summary": "Extended supervised speech token learning across semantic, prosodic and speaker tasks and introduced differentiable-reward post-training alongside a broader training corpus.",
      "title": "CosyVoice3: multitask speech tokenization and differentiable-reward post-training"
    },
    "record_id": "urn:uuid:28fd95b7-17c5-4e28-a175-7bc205368129",
    "version": 1,
    "source_ref": {
      "record_id": "urn:uuid:28fd95b7-17c5-4e28-a175-7bc205368129",
      "version": 1,
      "content_hash": "sha256:e395c9238b7b41546b52392023eaaf7715fbc59192713ca983f94f6907dd441b"
    },
    "export_view": "wrs-public-editorial-view-1",
    "public_view_hash": "sha256:8045710905c7e45461f32416ab6c8ede8b52d8945f0b5be11459824fa46c62bc"
  },
  "target": {
    "record_id": "urn:uuid:28fd95b7-17c5-4e28-a175-7bc205368129",
    "version": 1,
    "content_hash": "sha256:e395c9238b7b41546b52392023eaaf7715fbc59192713ca983f94f6907dd441b"
  },
  "published_grade": {
    "profile": "wrs-published-grade-1",
    "edition": "2026-10-28.1",
    "id": "du-et-al-cosyvoice3-multitask-tokenizer-reward-2025",
    "entry_id": "du-et-al-cosyvoice3-multitask-tokenizer-reward-2025",
    "revision": 1,
    "title": "CosyVoice3: multitask speech tokenization and differentiable-reward post-training",
    "context": "Speech tokenization and post-training",
    "track": "Human-designed research method or resource",
    "band": "B",
    "modifier": null,
    "status": "PROVISIONAL_EDITORIAL",
    "assessed_on": "2026-10-16",
    "evidence_cutoff": "2026-10-16",
    "achievement_state": "SUPPORTED_SCOPED_RESULT",
    "what_changed": "Extended supervised speech token learning across semantic, prosodic and speaker tasks and introduced differentiable-reward post-training alongside a broader training corpus.",
    "scope": "Extended supervised speech token learning across semantic, prosodic and speaker tasks and introduced differentiable-reward post-training alongside a broader training corpus. Inherits CosyVoice2 and multitask/reward learning. Acquired delta is specified tokenizer/post-training mechanism, not generic scaling of hours/parameters/languages. Reported multilingual quality is not universal robustness or independently validated human parity.",
    "evidence_summary": "The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Inherits CosyVoice2 and multitask/reward learning. Acquired delta is specified tokenizer/post-training mechanism, not generic scaling of hours/parameters/languages. Reported multilingual quality is not universal robustness or independently validated human parity.",
    "references": [
      {
        "title": "CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training",
        "url": "https://arxiv.org/abs/2505.17589v1",
        "locator": "Original contribution and stated assumptions, evaluation scope and limitations; source-linked assessment, not independent execution or proof certification."
      }
    ],
    "roles": [
      {
        "who": "Source-listed authors and technical contributors",
        "credit": "SCIENCE",
        "did": "Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities."
      }
    ],
    "target": {
      "record_id": "urn:uuid:28fd95b7-17c5-4e28-a175-7bc205368129",
      "version": 1,
      "content_hash": "sha256:e395c9238b7b41546b52392023eaaf7715fbc59192713ca983f94f6907dd441b"
    },
    "scientific_correctness_certified": false,
    "independent_expert_calibration": false,
    "method": "PROPRIETARY",
    "points": "55000"
  },
  "canonical_agent_submission": false,
  "scientific_verification_awarded": false,
  "reader_guide": null,
  "reading_status": "SOURCE_LINKED_NOTE"
}