{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:UYSY7OEGIRHBSOZTY6DXGTH5PF","short_pith_number":"pith:UYSY7OEG","schema_version":"1.0","canonical_sha256":"a6258fb886444e193b33c787734cfd79766d074e2d052b5259c377af931de2b5","source":{"kind":"arxiv","id":"2506.00843","version":1},"attestation_state":"computed","paper":{"title":"HASRD: Hierarchical Acoustic and Semantic Representation Disentanglement","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.SD"],"primary_cat":"eess.AS","authors_text":"Amir Hussein, Francois G. Germain, Gordon Wichern, Jonathan Le Roux, Sameer Khurana","submitted_at":"2025-06-01T05:38:39Z","abstract_excerpt":"Effective speech representations for spoken language models must balance semantic relevance with acoustic fidelity for high-quality reconstruction. However, existing approaches struggle to achieve both simultaneously. To address this, we introduce Hierarchical Acoustic and Semantic Representation Disentanglement (HASRD, pronounced `hazard'), a framework that factorizes self-supervised learning representations into discrete semantic and acoustic tokens. HASRD assigns the semantic representation to the first codebook, while encoding acoustic residuals in subsequent codebooks. This preserves ASR "},"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":"2506.00843","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"eess.AS","submitted_at":"2025-06-01T05:38:39Z","cross_cats_sorted":["cs.SD"],"title_canon_sha256":"d92c57d068affa0d66c75f728ecf6e67886aed2526dd4785dbf66de21e88c540","abstract_canon_sha256":"83d039de1195dab577834c49e8032381529dcdc97bbb62294628ea2772ab0448"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:13:41.510958Z","signature_b64":"6ZIQ6VvUvKUbjO708RG4Qxvu/2rX6JjTBDFfCoY0OZpede8vuUX/w0dJb8Z7gzfM2i+x02WUwBAEiGHBzz5ZDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"a6258fb886444e193b33c787734cfd79766d074e2d052b5259c377af931de2b5","last_reissued_at":"2026-07-05T11:13:41.510447Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:13:41.510447Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"HASRD: Hierarchical Acoustic and Semantic Representation Disentanglement","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.SD"],"primary_cat":"eess.AS","authors_text":"Amir Hussein, Francois G. Germain, Gordon Wichern, Jonathan Le Roux, Sameer Khurana","submitted_at":"2025-06-01T05:38:39Z","abstract_excerpt":"Effective speech representations for spoken language models must balance semantic relevance with acoustic fidelity for high-quality reconstruction. However, existing approaches struggle to achieve both simultaneously. To address this, we introduce Hierarchical Acoustic and Semantic Representation Disentanglement (HASRD, pronounced `hazard'), a framework that factorizes self-supervised learning representations into discrete semantic and acoustic tokens. HASRD assigns the semantic representation to the first codebook, while encoding acoustic residuals in subsequent codebooks. This preserves ASR "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2506.00843","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/2506.00843/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":"2506.00843","created_at":"2026-07-05T11:13:41.510511+00:00"},{"alias_kind":"arxiv_version","alias_value":"2506.00843v1","created_at":"2026-07-05T11:13:41.510511+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2506.00843","created_at":"2026-07-05T11:13:41.510511+00:00"},{"alias_kind":"pith_short_12","alias_value":"UYSY7OEGIRHB","created_at":"2026-07-05T11:13:41.510511+00:00"},{"alias_kind":"pith_short_16","alias_value":"UYSY7OEGIRHBSOZT","created_at":"2026-07-05T11:13:41.510511+00:00"},{"alias_kind":"pith_short_8","alias_value":"UYSY7OEG","created_at":"2026-07-05T11:13:41.510511+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.05927","citing_title":"Minimizing Modality Gap from the Input Side: Your Speech LLM Can Be a Prosody-Aware Text LLM","ref_index":40,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05927","citing_title":"Minimizing Modality Gap from the Input Side: Your Speech LLM Can Be a Prosody-Aware Text LLM","ref_index":40,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/UYSY7OEGIRHBSOZTY6DXGTH5PF","json":"https://pith.science/pith/UYSY7OEGIRHBSOZTY6DXGTH5PF.json","graph_json":"https://pith.science/api/pith-number/UYSY7OEGIRHBSOZTY6DXGTH5PF/graph.json","events_json":"https://pith.science/api/pith-number/UYSY7OEGIRHBSOZTY6DXGTH5PF/events.json","paper":"https://pith.science/paper/UYSY7OEG"},"agent_actions":{"view_html":"https://pith.science/pith/UYSY7OEGIRHBSOZTY6DXGTH5PF","download_json":"https://pith.science/pith/UYSY7OEGIRHBSOZTY6DXGTH5PF.json","view_paper":"https://pith.science/paper/UYSY7OEG","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2506.00843&json=true","fetch_graph":"https://pith.science/api/pith-number/UYSY7OEGIRHBSOZTY6DXGTH5PF/graph.json","fetch_events":"https://pith.science/api/pith-number/UYSY7OEGIRHBSOZTY6DXGTH5PF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/UYSY7OEGIRHBSOZTY6DXGTH5PF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/UYSY7OEGIRHBSOZTY6DXGTH5PF/action/storage_attestation","attest_author":"https://pith.science/pith/UYSY7OEGIRHBSOZTY6DXGTH5PF/action/author_attestation","sign_citation":"https://pith.science/pith/UYSY7OEGIRHBSOZTY6DXGTH5PF/action/citation_signature","submit_replication":"https://pith.science/pith/UYSY7OEGIRHBSOZTY6DXGTH5PF/action/replication_record"}},"created_at":"2026-07-05T11:13:41.510511+00:00","updated_at":"2026-07-05T11:13:41.510511+00:00"}