{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:LBBNJN3FG3LUTPBNVIBDK7FMEQ","short_pith_number":"pith:LBBNJN3F","schema_version":"1.0","canonical_sha256":"5842d4b76536d749bc2daa02357cac240bd1ad9054825edfc4b1baaddeab11b2","source":{"kind":"arxiv","id":"2506.11130","version":2},"attestation_state":"computed","paper":{"title":"A Self-Refining Framework for Enhancing ASR Using TTS-Synthesized Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.SD","eess.AS"],"primary_cat":"cs.CL","authors_text":"Chan-Jan Hsu, Cheng-Kang Chou, Ho-Lam Chung, Hsi-Chun Cheng, Hung-yi Lee, Kuan Po Huang, Liang-Hsuan Tseng, Yu-Kuan Fu","submitted_at":"2025-06-10T17:30:32Z","abstract_excerpt":"We propose a self-refining framework that enhances ASR performance with only unlabeled datasets. The process starts with an existing ASR model generating pseudo-labels on unannotated speech, which are then used to train a high-fidelity text-to-speech (TTS) system. Then, synthesized speech text pairs are bootstrapped into the original ASR system, completing the closed-loop self-improvement cycle. We demonstrated the effectiveness of the framework on Taiwanese Mandarin speech. Leveraging 6,000 hours of unlabeled speech, a moderate amount of text data, and synthetic content from the AI models, we"},"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":"2506.11130","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-06-10T17:30:32Z","cross_cats_sorted":["cs.AI","cs.SD","eess.AS"],"title_canon_sha256":"44e5b81ff8a573b1fb9aeb6b70597a89ba8c2ea19a227659033ad1cbb007021c","abstract_canon_sha256":"61f83d6c933f54c13a523770fb51a4f49c625a86bba6245f4b8bda3174802e30"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:22:19.098881Z","signature_b64":"yg4VXrpsIZVxIHTfnxV0GbTkfQh4Lo+cgXVG3WcXpWk0PBp4pUJ1EU9uoiBI1+1MqRe2ru7b0qQk2gfcFBkqDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5842d4b76536d749bc2daa02357cac240bd1ad9054825edfc4b1baaddeab11b2","last_reissued_at":"2026-07-05T11:22:19.098372Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:22:19.098372Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Self-Refining Framework for Enhancing ASR Using TTS-Synthesized Data","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.SD","eess.AS"],"primary_cat":"cs.CL","authors_text":"Chan-Jan Hsu, Cheng-Kang Chou, Ho-Lam Chung, Hsi-Chun Cheng, Hung-yi Lee, Kuan Po Huang, Liang-Hsuan Tseng, Yu-Kuan Fu","submitted_at":"2025-06-10T17:30:32Z","abstract_excerpt":"We propose a self-refining framework that enhances ASR performance with only unlabeled datasets. The process starts with an existing ASR model generating pseudo-labels on unannotated speech, which are then used to train a high-fidelity text-to-speech (TTS) system. Then, synthesized speech text pairs are bootstrapped into the original ASR system, completing the closed-loop self-improvement cycle. We demonstrated the effectiveness of the framework on Taiwanese Mandarin speech. Leveraging 6,000 hours of unlabeled speech, a moderate amount of text data, and synthetic content from the AI models, we"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.11130","kind":"arxiv","version":2},"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/2506.11130/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":"2506.11130","created_at":"2026-07-05T11:22:19.098428+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.11130v2","created_at":"2026-07-05T11:22:19.098428+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.11130","created_at":"2026-07-05T11:22:19.098428+00:00"},{"alias_kind":"pith_short_12","alias_value":"LBBNJN3FG3LU","created_at":"2026-07-05T11:22:19.098428+00:00"},{"alias_kind":"pith_short_16","alias_value":"LBBNJN3FG3LUTPBN","created_at":"2026-07-05T11:22:19.098428+00:00"},{"alias_kind":"pith_short_8","alias_value":"LBBNJN3F","created_at":"2026-07-05T11:22:19.098428+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":3,"sample":[{"citing_arxiv_id":"2607.08409","citing_title":"When Synthetic Speech Is All You Have: Better Call GRPO","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2607.06054","citing_title":"BlueMagpie-TTS: A Token-Efficient Tokenizer, Language Model, and TTS for Taiwanese-Accent Code-Switching Speech","ref_index":29,"is_internal_anchor":true},{"citing_arxiv_id":"2607.05364","citing_title":"REDDIT: Correcting Model-Generated Timestamp Drift in ASR without Forgetting via Replay-Based Distribution Editing","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2606.29031","citing_title":"How to Leverage Synthetic Speech for LLM-Based ASR Systems?","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10065","citing_title":"ASPIRin: Action Space Projection for Interactivity-Optimized Reinforcement Learning in Full-Duplex Speech Language Models","ref_index":15,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/LBBNJN3FG3LUTPBNVIBDK7FMEQ","json":"https://pith.science/pith/LBBNJN3FG3LUTPBNVIBDK7FMEQ.json","graph_json":"https://pith.science/api/pith-number/LBBNJN3FG3LUTPBNVIBDK7FMEQ/graph.json","events_json":"https://pith.science/api/pith-number/LBBNJN3FG3LUTPBNVIBDK7FMEQ/events.json","paper":"https://pith.science/paper/LBBNJN3F"},"agent_actions":{"view_html":"https://pith.science/pith/LBBNJN3FG3LUTPBNVIBDK7FMEQ","download_json":"https://pith.science/pith/LBBNJN3FG3LUTPBNVIBDK7FMEQ.json","view_paper":"https://pith.science/paper/LBBNJN3F","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.11130&json=true","fetch_graph":"https://pith.science/api/pith-number/LBBNJN3FG3LUTPBNVIBDK7FMEQ/graph.json","fetch_events":"https://pith.science/api/pith-number/LBBNJN3FG3LUTPBNVIBDK7FMEQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/LBBNJN3FG3LUTPBNVIBDK7FMEQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/LBBNJN3FG3LUTPBNVIBDK7FMEQ/action/storage_attestation","attest_author":"https://pith.science/pith/LBBNJN3FG3LUTPBNVIBDK7FMEQ/action/author_attestation","sign_citation":"https://pith.science/pith/LBBNJN3FG3LUTPBNVIBDK7FMEQ/action/citation_signature","submit_replication":"https://pith.science/pith/LBBNJN3FG3LUTPBNVIBDK7FMEQ/action/replication_record"}},"created_at":"2026-07-05T11:22:19.098428+00:00","updated_at":"2026-07-05T11:22:19.098428+00:00"}