{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:PIPADSK66J4CJZYA4ZEQCPQTFH","short_pith_number":"pith:PIPADSK6","schema_version":"1.0","canonical_sha256":"7a1e01c95ef27824e700e649013e1329ed8cd8ed32c12eb108afaa26043c1479","source":{"kind":"arxiv","id":"2504.02407","version":3},"attestation_state":"computed","paper":{"title":"F5R-TTS: Improving Flow-Matching based Text-to-Speech with Group Relative Policy Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.SD","authors_text":"Baoxun Wang, Bowen Wu, Jianye Mo, Qun Yu, Ruitong Xiao, Xiaohui Sun","submitted_at":"2025-04-03T08:57:15Z","abstract_excerpt":"We present F5R-TTS, a novel text-to-speech (TTS) system that integrates Group Relative Policy Optimization (GRPO) into a flow-matching based architecture. By reformulating the deterministic outputs of flow-matching TTS into probabilistic Gaussian distributions, our approach enables seamless integration of reinforcement learning algorithms. During pretraining, we train a probabilistically reformulated flow-matching based model which is derived from F5-TTS with an open-source dataset. In the subsequent reinforcement learning (RL) phase, we employ a GRPO-driven enhancement stage that leverages du"},"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":"2504.02407","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.SD","submitted_at":"2025-04-03T08:57:15Z","cross_cats_sorted":["eess.AS"],"title_canon_sha256":"4366bf121eb2dd0b68653c7c0de92e19fa60a71431d0c37be8a47f4721ba35df","abstract_canon_sha256":"95299c7753863860dc53a59c9453760eaf2cff76dadfad02f6d034221fa39477"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:52:21.731391Z","signature_b64":"U7DsrT15lWFfAJIJPHitn4etSAm+9ODzVC+yKhVDyIDTrJBOIsIbS3uRcEMwk23PxGK6TyowJI7JTsNucxNOAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"7a1e01c95ef27824e700e649013e1329ed8cd8ed32c12eb108afaa26043c1479","last_reissued_at":"2026-07-05T10:52:21.730849Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:52:21.730849Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"F5R-TTS: Improving Flow-Matching based Text-to-Speech with Group Relative Policy Optimization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.SD","authors_text":"Baoxun Wang, Bowen Wu, Jianye Mo, Qun Yu, Ruitong Xiao, Xiaohui Sun","submitted_at":"2025-04-03T08:57:15Z","abstract_excerpt":"We present F5R-TTS, a novel text-to-speech (TTS) system that integrates Group Relative Policy Optimization (GRPO) into a flow-matching based architecture. By reformulating the deterministic outputs of flow-matching TTS into probabilistic Gaussian distributions, our approach enables seamless integration of reinforcement learning algorithms. During pretraining, we train a probabilistically reformulated flow-matching based model which is derived from F5-TTS with an open-source dataset. In the subsequent reinforcement learning (RL) phase, we employ a GRPO-driven enhancement stage that leverages du"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.02407","kind":"arxiv","version":3},"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/2504.02407/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":"2504.02407","created_at":"2026-07-05T10:52:21.730903+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.02407v3","created_at":"2026-07-05T10:52:21.730903+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.02407","created_at":"2026-07-05T10:52:21.730903+00:00"},{"alias_kind":"pith_short_12","alias_value":"PIPADSK66J4C","created_at":"2026-07-05T10:52:21.730903+00:00"},{"alias_kind":"pith_short_16","alias_value":"PIPADSK66J4CJZYA","created_at":"2026-07-05T10:52:21.730903+00:00"},{"alias_kind":"pith_short_8","alias_value":"PIPADSK6","created_at":"2026-07-05T10:52:21.730903+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":9,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.23190","citing_title":"FlowTTS-GRPO: Online Reinforcement Learning with Multi-Objective Reward Optimization for Flow-Matching Based Text-to-Speech","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2606.09234","citing_title":"End-to-End Training for Discrete Token LLM based TTS System","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08602","citing_title":"Reinforcement Learning for Flow-Matching Policies with Density Transport","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05889","citing_title":"GLASS: GRPO-Trained LoRA for Acoustic Style Steering in Zero-Shot Text-to-Speech","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2607.00387","citing_title":"From Objectives to Applications: Aligning Architectural Biases in Audio Self-Supervised Learning","ref_index":139,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03455","citing_title":"WavTTS: Towards High-Quality Zero-Shot TTS via Direct Raw Waveform Modeling","ref_index":78,"is_internal_anchor":false},{"citing_arxiv_id":"2505.17589","citing_title":"CosyVoice 3: Towards In-the-wild Speech Generation via Scaling-up and Post-training","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2603.24936","citing_title":"TIGFlow-GRPO: Trajectory Forecasting via Interaction-Aware Flow Matching and Reward-Guided Optimization","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2505.05470","citing_title":"Flow-GRPO: Training Flow Matching Models via Online RL","ref_index":56,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/PIPADSK66J4CJZYA4ZEQCPQTFH","json":"https://pith.science/pith/PIPADSK66J4CJZYA4ZEQCPQTFH.json","graph_json":"https://pith.science/api/pith-number/PIPADSK66J4CJZYA4ZEQCPQTFH/graph.json","events_json":"https://pith.science/api/pith-number/PIPADSK66J4CJZYA4ZEQCPQTFH/events.json","paper":"https://pith.science/paper/PIPADSK6"},"agent_actions":{"view_html":"https://pith.science/pith/PIPADSK66J4CJZYA4ZEQCPQTFH","download_json":"https://pith.science/pith/PIPADSK66J4CJZYA4ZEQCPQTFH.json","view_paper":"https://pith.science/paper/PIPADSK6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.02407&json=true","fetch_graph":"https://pith.science/api/pith-number/PIPADSK66J4CJZYA4ZEQCPQTFH/graph.json","fetch_events":"https://pith.science/api/pith-number/PIPADSK66J4CJZYA4ZEQCPQTFH/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/PIPADSK66J4CJZYA4ZEQCPQTFH/action/timestamp_anchor","attest_storage":"https://pith.science/pith/PIPADSK66J4CJZYA4ZEQCPQTFH/action/storage_attestation","attest_author":"https://pith.science/pith/PIPADSK66J4CJZYA4ZEQCPQTFH/action/author_attestation","sign_citation":"https://pith.science/pith/PIPADSK66J4CJZYA4ZEQCPQTFH/action/citation_signature","submit_replication":"https://pith.science/pith/PIPADSK66J4CJZYA4ZEQCPQTFH/action/replication_record"}},"created_at":"2026-07-05T10:52:21.730903+00:00","updated_at":"2026-07-05T10:52:21.730903+00:00"}