{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:JSEKJ7PCGR3EKGLLZVV3NMYAFV","short_pith_number":"pith:JSEKJ7PC","schema_version":"1.0","canonical_sha256":"4c88a4fde2347645196bcd6bb6b3002d52154e47532b421536665aa7cd47d0db","source":{"kind":"arxiv","id":"2308.16415","version":1},"attestation_state":"computed","paper":{"title":"Knowledge Distillation from Non-streaming to Streaming ASR Encoder using Auxiliary Non-streaming Layer","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.CL","authors_text":"Jinkyu Lee, Kyuhong Shim, Kyuwoong Hwang, Simyung Chang","submitted_at":"2023-08-31T02:58:33Z","abstract_excerpt":"Streaming automatic speech recognition (ASR) models are restricted from accessing future context, which results in worse performance compared to the non-streaming models. To improve the performance of streaming ASR, knowledge distillation (KD) from the non-streaming to streaming model has been studied, mainly focusing on aligning the output token probabilities. In this paper, we propose a layer-to-layer KD from the teacher encoder to the student encoder. To ensure that features are extracted using the same context, we insert auxiliary non-streaming branches to the student and perform KD from t"},"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":"2308.16415","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2023-08-31T02:58:33Z","cross_cats_sorted":["eess.AS"],"title_canon_sha256":"f06aaf46e9e0abd53b7d18f9b77c88546ca730035c94f3d61ad86c2be7f4df7a","abstract_canon_sha256":"e68e857226df6bcbc5d59d7890ce1816c09ffe73d127ffe2456daee22a1c1852"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:46:27.433361Z","signature_b64":"Tyjgzd6p2Lq/nJjWcs3lf/p9vH9cr9dyB53ui6YixF9o880gLWwAqvM29HlggplwHIbz651KGQcEgwYohuXWCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"4c88a4fde2347645196bcd6bb6b3002d52154e47532b421536665aa7cd47d0db","last_reissued_at":"2026-07-05T06:46:27.432884Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:46:27.432884Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Knowledge Distillation from Non-streaming to Streaming ASR Encoder using Auxiliary Non-streaming Layer","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["eess.AS"],"primary_cat":"cs.CL","authors_text":"Jinkyu Lee, Kyuhong Shim, Kyuwoong Hwang, Simyung Chang","submitted_at":"2023-08-31T02:58:33Z","abstract_excerpt":"Streaming automatic speech recognition (ASR) models are restricted from accessing future context, which results in worse performance compared to the non-streaming models. To improve the performance of streaming ASR, knowledge distillation (KD) from the non-streaming to streaming model has been studied, mainly focusing on aligning the output token probabilities. In this paper, we propose a layer-to-layer KD from the teacher encoder to the student encoder. To ensure that features are extracted using the same context, we insert auxiliary non-streaming branches to the student and perform KD from t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2308.16415","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/2308.16415/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":"2308.16415","created_at":"2026-07-05T06:46:27.432938+00:00"},{"alias_kind":"arxiv_version","alias_value":"2308.16415v1","created_at":"2026-07-05T06:46:27.432938+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2308.16415","created_at":"2026-07-05T06:46:27.432938+00:00"},{"alias_kind":"pith_short_12","alias_value":"JSEKJ7PCGR3E","created_at":"2026-07-05T06:46:27.432938+00:00"},{"alias_kind":"pith_short_16","alias_value":"JSEKJ7PCGR3EKGLL","created_at":"2026-07-05T06:46:27.432938+00:00"},{"alias_kind":"pith_short_8","alias_value":"JSEKJ7PC","created_at":"2026-07-05T06:46:27.432938+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.22069","citing_title":"Delayed-KD: Delayed Knowledge Distillation based CTC for Low-Latency Streaming ASR","ref_index":15,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/JSEKJ7PCGR3EKGLLZVV3NMYAFV","json":"https://pith.science/pith/JSEKJ7PCGR3EKGLLZVV3NMYAFV.json","graph_json":"https://pith.science/api/pith-number/JSEKJ7PCGR3EKGLLZVV3NMYAFV/graph.json","events_json":"https://pith.science/api/pith-number/JSEKJ7PCGR3EKGLLZVV3NMYAFV/events.json","paper":"https://pith.science/paper/JSEKJ7PC"},"agent_actions":{"view_html":"https://pith.science/pith/JSEKJ7PCGR3EKGLLZVV3NMYAFV","download_json":"https://pith.science/pith/JSEKJ7PCGR3EKGLLZVV3NMYAFV.json","view_paper":"https://pith.science/paper/JSEKJ7PC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2308.16415&json=true","fetch_graph":"https://pith.science/api/pith-number/JSEKJ7PCGR3EKGLLZVV3NMYAFV/graph.json","fetch_events":"https://pith.science/api/pith-number/JSEKJ7PCGR3EKGLLZVV3NMYAFV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/JSEKJ7PCGR3EKGLLZVV3NMYAFV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/JSEKJ7PCGR3EKGLLZVV3NMYAFV/action/storage_attestation","attest_author":"https://pith.science/pith/JSEKJ7PCGR3EKGLLZVV3NMYAFV/action/author_attestation","sign_citation":"https://pith.science/pith/JSEKJ7PCGR3EKGLLZVV3NMYAFV/action/citation_signature","submit_replication":"https://pith.science/pith/JSEKJ7PCGR3EKGLLZVV3NMYAFV/action/replication_record"}},"created_at":"2026-07-05T06:46:27.432938+00:00","updated_at":"2026-07-05T06:46:27.432938+00:00"}