{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:6P3ZN4HOOKXMZ75VTISTPSNJ2P","short_pith_number":"pith:6P3ZN4HO","schema_version":"1.0","canonical_sha256":"f3f796f0ee72aeccffb59a2537c9a9d3d176827a2ac855894639ef39ea9f0e86","source":{"kind":"arxiv","id":"2302.12757","version":1},"attestation_state":"computed","paper":{"title":"Ensemble knowledge distillation of self-supervised speech models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.SD"],"primary_cat":"eess.AS","authors_text":"Hung-yi Lee, Kai-Wei Chang, Kuan-Po Huang, Po-Chieh Yen, Tsu-Yuan Hsu, Tzu-hsun Feng, Wei-Cheng Tseng, Yu-Kuan Fu","submitted_at":"2023-02-24T17:15:39Z","abstract_excerpt":"Distilled self-supervised models have shown competitive performance and efficiency in recent years. However, there is a lack of experience in jointly distilling multiple self-supervised speech models. In our work, we performed Ensemble Knowledge Distillation (EKD) on various self-supervised speech models such as HuBERT, RobustHuBERT, and WavLM. We tried two different aggregation techniques, layerwise-average and layerwise-concatenation, to the representations of different teacher models and found that the former was more effective. On top of that, we proposed a multiple prediction head method "},"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":"2302.12757","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"eess.AS","submitted_at":"2023-02-24T17:15:39Z","cross_cats_sorted":["cs.CL","cs.SD"],"title_canon_sha256":"36e310a01bd8464537f51f3c8c7327fca6e971a5a3473b2d20ec1a764e162a71","abstract_canon_sha256":"45d5b92ff49d8ab0babf45dcfbd504c2c4de932e7b4ff927fa8fdbb0589d20c3"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T05:45:20.953995Z","signature_b64":"5Z5RYtXKxkmgq4AU6ZXEJvJ+Ty9Yqy6okoo2csYBVgE4WqvZUvFrMs+z81ifMk6H3Bue8K7Qaiz98EgCHHX9DQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f3f796f0ee72aeccffb59a2537c9a9d3d176827a2ac855894639ef39ea9f0e86","last_reissued_at":"2026-07-05T05:45:20.953660Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T05:45:20.953660Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Ensemble knowledge distillation of self-supervised speech models","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.CL","cs.SD"],"primary_cat":"eess.AS","authors_text":"Hung-yi Lee, Kai-Wei Chang, Kuan-Po Huang, Po-Chieh Yen, Tsu-Yuan Hsu, Tzu-hsun Feng, Wei-Cheng Tseng, Yu-Kuan Fu","submitted_at":"2023-02-24T17:15:39Z","abstract_excerpt":"Distilled self-supervised models have shown competitive performance and efficiency in recent years. However, there is a lack of experience in jointly distilling multiple self-supervised speech models. In our work, we performed Ensemble Knowledge Distillation (EKD) on various self-supervised speech models such as HuBERT, RobustHuBERT, and WavLM. We tried two different aggregation techniques, layerwise-average and layerwise-concatenation, to the representations of different teacher models and found that the former was more effective. On top of that, we proposed a multiple prediction head method "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.12757","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/2302.12757/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":"2302.12757","created_at":"2026-07-05T05:45:20.953721+00:00"},{"alias_kind":"arxiv_version","alias_value":"2302.12757v1","created_at":"2026-07-05T05:45:20.953721+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.12757","created_at":"2026-07-05T05:45:20.953721+00:00"},{"alias_kind":"pith_short_12","alias_value":"6P3ZN4HOOKXM","created_at":"2026-07-05T05:45:20.953721+00:00"},{"alias_kind":"pith_short_16","alias_value":"6P3ZN4HOOKXMZ75V","created_at":"2026-07-05T05:45:20.953721+00:00"},{"alias_kind":"pith_short_8","alias_value":"6P3ZN4HO","created_at":"2026-07-05T05:45:20.953721+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6P3ZN4HOOKXMZ75VTISTPSNJ2P","json":"https://pith.science/pith/6P3ZN4HOOKXMZ75VTISTPSNJ2P.json","graph_json":"https://pith.science/api/pith-number/6P3ZN4HOOKXMZ75VTISTPSNJ2P/graph.json","events_json":"https://pith.science/api/pith-number/6P3ZN4HOOKXMZ75VTISTPSNJ2P/events.json","paper":"https://pith.science/paper/6P3ZN4HO"},"agent_actions":{"view_html":"https://pith.science/pith/6P3ZN4HOOKXMZ75VTISTPSNJ2P","download_json":"https://pith.science/pith/6P3ZN4HOOKXMZ75VTISTPSNJ2P.json","view_paper":"https://pith.science/paper/6P3ZN4HO","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2302.12757&json=true","fetch_graph":"https://pith.science/api/pith-number/6P3ZN4HOOKXMZ75VTISTPSNJ2P/graph.json","fetch_events":"https://pith.science/api/pith-number/6P3ZN4HOOKXMZ75VTISTPSNJ2P/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6P3ZN4HOOKXMZ75VTISTPSNJ2P/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6P3ZN4HOOKXMZ75VTISTPSNJ2P/action/storage_attestation","attest_author":"https://pith.science/pith/6P3ZN4HOOKXMZ75VTISTPSNJ2P/action/author_attestation","sign_citation":"https://pith.science/pith/6P3ZN4HOOKXMZ75VTISTPSNJ2P/action/citation_signature","submit_replication":"https://pith.science/pith/6P3ZN4HOOKXMZ75VTISTPSNJ2P/action/replication_record"}},"created_at":"2026-07-05T05:45:20.953721+00:00","updated_at":"2026-07-05T05:45:20.953721+00:00"}