{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:3OOLJGNDLYPTP6QE3XVXO7VFT3","short_pith_number":"pith:3OOLJGND","schema_version":"1.0","canonical_sha256":"db9cb499a35e1f37fa04ddeb777ea59efed2d019f9b234e781338882f2ca9f51","source":{"kind":"arxiv","id":"2410.02597","version":3},"attestation_state":"computed","paper":{"title":"HAINAN: Fast and Accurate Transducer for Hybrid-Autoregressive ASR","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Boris Ginsburg, Hainan Xu, Travis M. Bartley, Vladimir Bataev","submitted_at":"2024-10-03T15:38:20Z","abstract_excerpt":"We present Hybrid-Autoregressive INference TrANsducers (HAINAN), a novel architecture for speech recognition that extends the Token-and-Duration Transducer (TDT) model. Trained with randomly masked predictor network outputs, HAINAN supports both autoregressive inference with all network components and non-autoregressive inference without the predictor. Additionally, we propose a novel semi-autoregressive inference paradigm that first generates an initial hypothesis using non-autoregressive inference, followed by refinement steps where each token prediction is regenerated using parallelized aut"},"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":"2410.02597","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-10-03T15:38:20Z","cross_cats_sorted":[],"title_canon_sha256":"d804ef7af4e99101a626848383d8cc3be4c866f725f8735129be08761a7248f7","abstract_canon_sha256":"b7d39d14d777e06fc423af0616cd0fb3554c3adf00ee57a6f93704db459dc8c3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:58:18.984801Z","signature_b64":"u28Fdy0h/msXQ4Hmvvt1ZQ9ftZaT3W20EVfxvt6J76YmkQZ6dQxtNgYoyMeNRySni4WFlfYQLAFs1XO/PKOPBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"db9cb499a35e1f37fa04ddeb777ea59efed2d019f9b234e781338882f2ca9f51","last_reissued_at":"2026-07-05T10:58:18.984280Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:58:18.984280Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"HAINAN: Fast and Accurate Transducer for Hybrid-Autoregressive ASR","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Boris Ginsburg, Hainan Xu, Travis M. Bartley, Vladimir Bataev","submitted_at":"2024-10-03T15:38:20Z","abstract_excerpt":"We present Hybrid-Autoregressive INference TrANsducers (HAINAN), a novel architecture for speech recognition that extends the Token-and-Duration Transducer (TDT) model. Trained with randomly masked predictor network outputs, HAINAN supports both autoregressive inference with all network components and non-autoregressive inference without the predictor. Additionally, we propose a novel semi-autoregressive inference paradigm that first generates an initial hypothesis using non-autoregressive inference, followed by refinement steps where each token prediction is regenerated using parallelized aut"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.02597","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/2410.02597/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":"2410.02597","created_at":"2026-07-05T10:58:18.984371+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.02597v3","created_at":"2026-07-05T10:58:18.984371+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.02597","created_at":"2026-07-05T10:58:18.984371+00:00"},{"alias_kind":"pith_short_12","alias_value":"3OOLJGNDLYPT","created_at":"2026-07-05T10:58:18.984371+00:00"},{"alias_kind":"pith_short_16","alias_value":"3OOLJGNDLYPTP6QE","created_at":"2026-07-05T10:58:18.984371+00:00"},{"alias_kind":"pith_short_8","alias_value":"3OOLJGND","created_at":"2026-07-05T10:58:18.984371+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.13765","citing_title":"WIND: Accelerated RNN-T Decoding with Windowed Inference for Non-blank Detection","ref_index":31,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3OOLJGNDLYPTP6QE3XVXO7VFT3","json":"https://pith.science/pith/3OOLJGNDLYPTP6QE3XVXO7VFT3.json","graph_json":"https://pith.science/api/pith-number/3OOLJGNDLYPTP6QE3XVXO7VFT3/graph.json","events_json":"https://pith.science/api/pith-number/3OOLJGNDLYPTP6QE3XVXO7VFT3/events.json","paper":"https://pith.science/paper/3OOLJGND"},"agent_actions":{"view_html":"https://pith.science/pith/3OOLJGNDLYPTP6QE3XVXO7VFT3","download_json":"https://pith.science/pith/3OOLJGNDLYPTP6QE3XVXO7VFT3.json","view_paper":"https://pith.science/paper/3OOLJGND","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.02597&json=true","fetch_graph":"https://pith.science/api/pith-number/3OOLJGNDLYPTP6QE3XVXO7VFT3/graph.json","fetch_events":"https://pith.science/api/pith-number/3OOLJGNDLYPTP6QE3XVXO7VFT3/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3OOLJGNDLYPTP6QE3XVXO7VFT3/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3OOLJGNDLYPTP6QE3XVXO7VFT3/action/storage_attestation","attest_author":"https://pith.science/pith/3OOLJGNDLYPTP6QE3XVXO7VFT3/action/author_attestation","sign_citation":"https://pith.science/pith/3OOLJGNDLYPTP6QE3XVXO7VFT3/action/citation_signature","submit_replication":"https://pith.science/pith/3OOLJGNDLYPTP6QE3XVXO7VFT3/action/replication_record"}},"created_at":"2026-07-05T10:58:18.984371+00:00","updated_at":"2026-07-05T10:58:18.984371+00:00"}