{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:PBEMVMG4OCHVF2AILDN2QRI7RG","short_pith_number":"pith:PBEMVMG4","schema_version":"1.0","canonical_sha256":"7848cab0dc708f52e80858dba8451f899f1ce1524624768c8dfded3a869c554b","source":{"kind":"arxiv","id":"2508.19205","version":1},"attestation_state":"computed","paper":{"title":"VibeVoice Technical Report","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.SD","eess.AS"],"primary_cat":"cs.CL","authors_text":"Furu Wei, Hangbo Bao, Jianwei Yu, Li Dong, Shaohan Huang, Weijiang Xu, Wenhui Wang, Yan Xia, Yaoyao Chang, Yi Zhu, Yutao Sun, Zehua Wang, Zhiliang Peng","submitted_at":"2025-08-26T17:09:12Z","abstract_excerpt":"This report presents VibeVoice, a novel model designed to synthesize long-form speech with multiple speakers by employing next-token diffusion, which is a unified method for modeling continuous data by autoregressively generating latent vectors via diffusion. To enable this, we introduce a novel continuous speech tokenizer that, when compared to the popular Encodec model, improves data compression by 80 times while maintaining comparable performance. The tokenizer effectively preserves audio fidelity while significantly boosting computational efficiency for processing long sequences. Thus, Vib"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2508.19205","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-08-26T17:09:12Z","cross_cats_sorted":["cs.AI","cs.SD","eess.AS"],"title_canon_sha256":"9cb3da66ead2639341f6fff10ea0d291121fceab292e9cb0ae61b19b76eeff17","abstract_canon_sha256":"b67c64fc69bcf6a0ce5cf9bae105ec64d4799f230b9f3b99f988b5c94f18cd3e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:59:39.612102Z","signature_b64":"tZwYSMI8kYYNG9cZbv9mjL/7VGOjy34JOFYg9vVlf0wWYT4/lrdqH4zQZHUwkk4BlSmsefOuMoUjzlJcsX/aCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7848cab0dc708f52e80858dba8451f899f1ce1524624768c8dfded3a869c554b","last_reissued_at":"2026-07-05T11:59:39.611732Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:59:39.611732Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"VibeVoice Technical Report","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.SD","eess.AS"],"primary_cat":"cs.CL","authors_text":"Furu Wei, Hangbo Bao, Jianwei Yu, Li Dong, Shaohan Huang, Weijiang Xu, Wenhui Wang, Yan Xia, Yaoyao Chang, Yi Zhu, Yutao Sun, Zehua Wang, Zhiliang Peng","submitted_at":"2025-08-26T17:09:12Z","abstract_excerpt":"This report presents VibeVoice, a novel model designed to synthesize long-form speech with multiple speakers by employing next-token diffusion, which is a unified method for modeling continuous data by autoregressively generating latent vectors via diffusion. To enable this, we introduce a novel continuous speech tokenizer that, when compared to the popular Encodec model, improves data compression by 80 times while maintaining comparable performance. The tokenizer effectively preserves audio fidelity while significantly boosting computational efficiency for processing long sequences. Thus, Vib"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.19205","kind":"arxiv","version":1},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2508.19205/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2508.19205","created_at":"2026-07-05T11:59:39.611782+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.19205v1","created_at":"2026-07-05T11:59:39.611782+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.19205","created_at":"2026-07-05T11:59:39.611782+00:00"},{"alias_kind":"pith_short_12","alias_value":"PBEMVMG4OCHV","created_at":"2026-07-05T11:59:39.611782+00:00"},{"alias_kind":"pith_short_16","alias_value":"PBEMVMG4OCHVF2AI","created_at":"2026-07-05T11:59:39.611782+00:00"},{"alias_kind":"pith_short_8","alias_value":"PBEMVMG4","created_at":"2026-07-05T11:59:39.611782+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":17,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.10231","citing_title":"LLM can Read Spectrogram: Encoder-free Speech-Language Modeling","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2606.18485","citing_title":"MagpieTTS-LF: Inference-Time Long-Form Speech Generation Without Training on Long-Form data","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2606.09098","citing_title":"HoliDubber: Holistic Video Dubbing for Complex Acoustic Scenes via Text-Guided Audio Synthesis","ref_index":53,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07080","citing_title":"dots.tts Technical Report","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06928","citing_title":"VoxCPM2 Technical Report","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06357","citing_title":"F3-Tokenizer: Taming Audio Autoencoder Latents for Understanding and Generation","ref_index":12,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30993","citing_title":"SwanVoice: Expressive Long-Form Zero-Shot Speech Synthesis for Both Monologue and Dialogue","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2605.25659","citing_title":"StreamChar: Long-Horizon Streaming Character Audio-Video Generation with Decoupled Orchestration","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27740","citing_title":"UNIQUE: Universal Top-k Sparse Attention for Training-free Inference and Sparsity-aware Training","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27258","citing_title":"PilotTTS: A Disciplined Modular Recipe for Competitive Speech Synthesis","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30993","citing_title":"SwanVoice: Expressive Long-Form Zero-Shot Speech Synthesis for Both Monologue and Dialogue","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2606.04418","citing_title":"CleanCodec: Efficient and Robust Speech Tokenization via Perceptually Guided Encoding","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2512.14234","citing_title":"ViBES: A Conversational Agent with Behaviorally-Intelligent 3D Virtual Body","ref_index":88,"is_internal_anchor":false},{"citing_arxiv_id":"2601.15621","citing_title":"Qwen3-TTS Technical Report","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12383","citing_title":"On the Distillation Loss Functions of Speech VAE for Unified Reconstruction, Understanding, and Generation","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17958","citing_title":"MINT-Bench: A Comprehensive Multilingual Benchmark for Instruction-Following Text-to-Speech","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2604.20267","citing_title":"ATIR: Towards Audio-Text Interleaved Contextual Retrieval","ref_index":36,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PBEMVMG4OCHVF2AILDN2QRI7RG","json":"https://pith.science/pith/PBEMVMG4OCHVF2AILDN2QRI7RG.json","graph_json":"https://pith.science/api/pith-number/PBEMVMG4OCHVF2AILDN2QRI7RG/graph.json","events_json":"https://pith.science/api/pith-number/PBEMVMG4OCHVF2AILDN2QRI7RG/events.json","paper":"https://pith.science/paper/PBEMVMG4"},"agent_actions":{"view_html":"https://pith.science/pith/PBEMVMG4OCHVF2AILDN2QRI7RG","download_json":"https://pith.science/pith/PBEMVMG4OCHVF2AILDN2QRI7RG.json","view_paper":"https://pith.science/paper/PBEMVMG4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.19205&json=true","fetch_graph":"https://pith.science/api/pith-number/PBEMVMG4OCHVF2AILDN2QRI7RG/graph.json","fetch_events":"https://pith.science/api/pith-number/PBEMVMG4OCHVF2AILDN2QRI7RG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PBEMVMG4OCHVF2AILDN2QRI7RG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PBEMVMG4OCHVF2AILDN2QRI7RG/action/storage_attestation","attest_author":"https://pith.science/pith/PBEMVMG4OCHVF2AILDN2QRI7RG/action/author_attestation","sign_citation":"https://pith.science/pith/PBEMVMG4OCHVF2AILDN2QRI7RG/action/citation_signature","submit_replication":"https://pith.science/pith/PBEMVMG4OCHVF2AILDN2QRI7RG/action/replication_record"}},"created_at":"2026-07-05T11:59:39.611782+00:00","updated_at":"2026-07-05T11:59:39.611782+00:00"}