{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2020:J5QWS3EEWB6PEXI2RY5LVK4ITN","short_pith_number":"pith:J5QWS3EE","schema_version":"1.0","canonical_sha256":"4f61696c84b07cf25d1a8e3abaab889b487392198c1d01a2db8fdaace6e21df6","source":{"kind":"arxiv","id":"2008.03822","version":1},"attestation_state":"computed","paper":{"title":"Distilling the Knowledge of BERT for Sequence-to-Sequence ASR","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.CL","authors_text":"Hayato Futami, Hirofumi Inaguma, Masato Mimura, Sei Ueno, Shinsuke Sakai, Tatsuya Kawahara","submitted_at":"2020-08-09T21:48:02Z","abstract_excerpt":"Attention-based sequence-to-sequence (seq2seq) models have achieved promising results in automatic speech recognition (ASR). However, as these models decode in a left-to-right way, they do not have access to context on the right. We leverage both left and right context by applying BERT as an external language model to seq2seq ASR through knowledge distillation. In our proposed method, BERT generates soft labels to guide the training of seq2seq ASR. Furthermore, we leverage context beyond the current utterance as input to BERT. Experimental evaluations show that our method significantly improve"},"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":"2008.03822","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2020-08-09T21:48:02Z","cross_cats_sorted":["eess.AS"],"title_canon_sha256":"5d7924f06b5485cf6eb288de229e0b15e1c6f4a947b2246b3d912625e941ee7a","abstract_canon_sha256":"08dbf293bf2e6e8231d966a910af1a338f93c9cf80aa5b091d16ed8f5fa85293"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:25:45.777298Z","signature_b64":"KNXNrhSde1G4wepc8jw+L0d50TupYDeG1obk4t71Yl7bZRqdJMEWDOfxfF/p4DeamxEg9B2+6vG6rhVFuvaVDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4f61696c84b07cf25d1a8e3abaab889b487392198c1d01a2db8fdaace6e21df6","last_reissued_at":"2026-07-05T01:25:45.776837Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:25:45.776837Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Distilling the Knowledge of BERT for Sequence-to-Sequence ASR","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.CL","authors_text":"Hayato Futami, Hirofumi Inaguma, Masato Mimura, Sei Ueno, Shinsuke Sakai, Tatsuya Kawahara","submitted_at":"2020-08-09T21:48:02Z","abstract_excerpt":"Attention-based sequence-to-sequence (seq2seq) models have achieved promising results in automatic speech recognition (ASR). However, as these models decode in a left-to-right way, they do not have access to context on the right. We leverage both left and right context by applying BERT as an external language model to seq2seq ASR through knowledge distillation. In our proposed method, BERT generates soft labels to guide the training of seq2seq ASR. Furthermore, we leverage context beyond the current utterance as input to BERT. Experimental evaluations show that our method significantly improve"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2008.03822","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/2008.03822/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":"2008.03822","created_at":"2026-07-05T01:25:45.776892+00:00"},{"alias_kind":"arxiv_version","alias_value":"2008.03822v1","created_at":"2026-07-05T01:25:45.776892+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2008.03822","created_at":"2026-07-05T01:25:45.776892+00:00"},{"alias_kind":"pith_short_12","alias_value":"J5QWS3EEWB6P","created_at":"2026-07-05T01:25:45.776892+00:00"},{"alias_kind":"pith_short_16","alias_value":"J5QWS3EEWB6PEXI2","created_at":"2026-07-05T01:25:45.776892+00:00"},{"alias_kind":"pith_short_8","alias_value":"J5QWS3EE","created_at":"2026-07-05T01:25:45.776892+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2508.07285","citing_title":"Non-Intrusive Automatic Speech Recognition Refinement: A Survey","ref_index":20,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/J5QWS3EEWB6PEXI2RY5LVK4ITN","json":"https://pith.science/pith/J5QWS3EEWB6PEXI2RY5LVK4ITN.json","graph_json":"https://pith.science/api/pith-number/J5QWS3EEWB6PEXI2RY5LVK4ITN/graph.json","events_json":"https://pith.science/api/pith-number/J5QWS3EEWB6PEXI2RY5LVK4ITN/events.json","paper":"https://pith.science/paper/J5QWS3EE"},"agent_actions":{"view_html":"https://pith.science/pith/J5QWS3EEWB6PEXI2RY5LVK4ITN","download_json":"https://pith.science/pith/J5QWS3EEWB6PEXI2RY5LVK4ITN.json","view_paper":"https://pith.science/paper/J5QWS3EE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2008.03822&json=true","fetch_graph":"https://pith.science/api/pith-number/J5QWS3EEWB6PEXI2RY5LVK4ITN/graph.json","fetch_events":"https://pith.science/api/pith-number/J5QWS3EEWB6PEXI2RY5LVK4ITN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/J5QWS3EEWB6PEXI2RY5LVK4ITN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/J5QWS3EEWB6PEXI2RY5LVK4ITN/action/storage_attestation","attest_author":"https://pith.science/pith/J5QWS3EEWB6PEXI2RY5LVK4ITN/action/author_attestation","sign_citation":"https://pith.science/pith/J5QWS3EEWB6PEXI2RY5LVK4ITN/action/citation_signature","submit_replication":"https://pith.science/pith/J5QWS3EEWB6PEXI2RY5LVK4ITN/action/replication_record"}},"created_at":"2026-07-05T01:25:45.776892+00:00","updated_at":"2026-07-05T01:25:45.776892+00:00"}