{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:IPN2BR5W5E6C3XNNANINTNTMUB","short_pith_number":"pith:IPN2BR5W","schema_version":"1.0","canonical_sha256":"43dba0c7b6e93c2dddad0350d9b66ca06b6513feb4b2011ac520250c6a5a0407","source":{"kind":"arxiv","id":"2501.04521","version":2},"attestation_state":"computed","paper":{"title":"Right Label Context in End-to-End Training of Time-Synchronous ASR Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.SD","authors_text":"Hermann Ney, Ralf Schl\\\"uter, Tina Raissi","submitted_at":"2025-01-08T14:14:19Z","abstract_excerpt":"Current time-synchronous sequence-to-sequence automatic speech recognition (ASR) models are trained by using sequence level cross-entropy that sums over all alignments. Due to the discriminative formulation, incorporating the right label context into the training criterion's gradient causes normalization problems and is not mathematically well-defined. The classic hybrid neural network hidden Markov model (NN-HMM) with its inherent generative formulation enables conditioning on the right label context. However, due to the HMM state-tying the identity of the right label context is never modeled"},"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":"2501.04521","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SD","submitted_at":"2025-01-08T14:14:19Z","cross_cats_sorted":["eess.AS"],"title_canon_sha256":"655a019a3157ae11dde82635e0b3d80b7fd7d74b4019b96f4e2b986fb73f9cd1","abstract_canon_sha256":"a11f72e7ab2b24917a2c9017da88b3a844798171777fc42f439d534e34aab860"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:59:02.700370Z","signature_b64":"fCGQQvUXJpoykhnS7o2lhdKnYQNjBwdllHs1psYgt49x4v35LI+FzQY5nV/4ffbLL5U7q5+dV5z09Dck6lzVDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"43dba0c7b6e93c2dddad0350d9b66ca06b6513feb4b2011ac520250c6a5a0407","last_reissued_at":"2026-07-05T09:59:02.700006Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:59:02.700006Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Right Label Context in End-to-End Training of Time-Synchronous ASR Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.SD","authors_text":"Hermann Ney, Ralf Schl\\\"uter, Tina Raissi","submitted_at":"2025-01-08T14:14:19Z","abstract_excerpt":"Current time-synchronous sequence-to-sequence automatic speech recognition (ASR) models are trained by using sequence level cross-entropy that sums over all alignments. Due to the discriminative formulation, incorporating the right label context into the training criterion's gradient causes normalization problems and is not mathematically well-defined. The classic hybrid neural network hidden Markov model (NN-HMM) with its inherent generative formulation enables conditioning on the right label context. However, due to the HMM state-tying the identity of the right label context is never modeled"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.04521","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/2501.04521/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":"2501.04521","created_at":"2026-07-05T09:59:02.700062+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.04521v2","created_at":"2026-07-05T09:59:02.700062+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.04521","created_at":"2026-07-05T09:59:02.700062+00:00"},{"alias_kind":"pith_short_12","alias_value":"IPN2BR5W5E6C","created_at":"2026-07-05T09:59:02.700062+00:00"},{"alias_kind":"pith_short_16","alias_value":"IPN2BR5W5E6C3XNN","created_at":"2026-07-05T09:59:02.700062+00:00"},{"alias_kind":"pith_short_8","alias_value":"IPN2BR5W","created_at":"2026-07-05T09:59:02.700062+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.06831","citing_title":"Gradient-Based Speech-to-Text Alignment for Any ASR Model: From CTC to Speech LLMs","ref_index":24,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IPN2BR5W5E6C3XNNANINTNTMUB","json":"https://pith.science/pith/IPN2BR5W5E6C3XNNANINTNTMUB.json","graph_json":"https://pith.science/api/pith-number/IPN2BR5W5E6C3XNNANINTNTMUB/graph.json","events_json":"https://pith.science/api/pith-number/IPN2BR5W5E6C3XNNANINTNTMUB/events.json","paper":"https://pith.science/paper/IPN2BR5W"},"agent_actions":{"view_html":"https://pith.science/pith/IPN2BR5W5E6C3XNNANINTNTMUB","download_json":"https://pith.science/pith/IPN2BR5W5E6C3XNNANINTNTMUB.json","view_paper":"https://pith.science/paper/IPN2BR5W","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.04521&json=true","fetch_graph":"https://pith.science/api/pith-number/IPN2BR5W5E6C3XNNANINTNTMUB/graph.json","fetch_events":"https://pith.science/api/pith-number/IPN2BR5W5E6C3XNNANINTNTMUB/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IPN2BR5W5E6C3XNNANINTNTMUB/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IPN2BR5W5E6C3XNNANINTNTMUB/action/storage_attestation","attest_author":"https://pith.science/pith/IPN2BR5W5E6C3XNNANINTNTMUB/action/author_attestation","sign_citation":"https://pith.science/pith/IPN2BR5W5E6C3XNNANINTNTMUB/action/citation_signature","submit_replication":"https://pith.science/pith/IPN2BR5W5E6C3XNNANINTNTMUB/action/replication_record"}},"created_at":"2026-07-05T09:59:02.700062+00:00","updated_at":"2026-07-05T09:59:02.700062+00:00"}