{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:DT5KZ6ZIJPQS2D6HL5I4RKGEE4","short_pith_number":"pith:DT5KZ6ZI","schema_version":"1.0","canonical_sha256":"1cfaacfb284be12d0fc75f51c8a8c4270f2da9fe83b583b745d7f626e3e24dd1","source":{"kind":"arxiv","id":"2508.14623","version":2},"attestation_state":"computed","paper":{"title":"A Study of the Scale Invariant Signal to Distortion Ratio in Speech Separation with Noisy References","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.SD"],"primary_cat":"eess.AS","authors_text":"Jesper Rindom Jensen, Mads Gr{\\ae}sb{\\o}ll Christensen, Simon Dahl Jepsen","submitted_at":"2025-08-20T11:22:11Z","abstract_excerpt":"This paper examines the implications of using the Scale-Invariant Signal-to-Distortion Ratio (SI-SDR) as both evaluation and training objective in supervised speech separation, when the training references contain noise, as is the case with the de facto benchmark WSJ0-2Mix. A derivation of the SI-SDR with noisy references reveals that noise limits the achievable SI-SDR, or leads to undesired noise in the separated outputs. To address this, a method is proposed to enhance references and augment the mixtures with WHAM!, aiming to train models that avoid learning noisy references. Two models trai"},"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":"2508.14623","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"eess.AS","submitted_at":"2025-08-20T11:22:11Z","cross_cats_sorted":["cs.AI","cs.SD"],"title_canon_sha256":"0755d3ecde8f1dd11e7de7a998a92dae8483bd73bd72fbc90753d6cacd42246d","abstract_canon_sha256":"907dd815f93126f96bbe854c60bd28580e85baaef51504525d205ed552b1d96a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-04T01:08:31.169524Z","signature_b64":"azoY3zfWDdfhE+aIZOrxFGmm+R2RXR7h/ZzFU0/iSebobggY5L8bTplKPLn9k1ZgAIAGqhsN+B8PxB9gP3dsAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1cfaacfb284be12d0fc75f51c8a8c4270f2da9fe83b583b745d7f626e3e24dd1","last_reissued_at":"2026-06-04T01:08:31.168536Z","signature_status":"signed_v1","first_computed_at":"2026-06-04T01:08:31.168536Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Study of the Scale Invariant Signal to Distortion Ratio in Speech Separation with Noisy References","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.SD"],"primary_cat":"eess.AS","authors_text":"Jesper Rindom Jensen, Mads Gr{\\ae}sb{\\o}ll Christensen, Simon Dahl Jepsen","submitted_at":"2025-08-20T11:22:11Z","abstract_excerpt":"This paper examines the implications of using the Scale-Invariant Signal-to-Distortion Ratio (SI-SDR) as both evaluation and training objective in supervised speech separation, when the training references contain noise, as is the case with the de facto benchmark WSJ0-2Mix. A derivation of the SI-SDR with noisy references reveals that noise limits the achievable SI-SDR, or leads to undesired noise in the separated outputs. To address this, a method is proposed to enhance references and augment the mixtures with WHAM!, aiming to train models that avoid learning noisy references. Two models trai"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.14623","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/2508.14623/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":"2508.14623","created_at":"2026-06-04T01:08:31.168681+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.14623v2","created_at":"2026-06-04T01:08:31.168681+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.14623","created_at":"2026-06-04T01:08:31.168681+00:00"},{"alias_kind":"pith_short_12","alias_value":"DT5KZ6ZIJPQS","created_at":"2026-06-04T01:08:31.168681+00:00"},{"alias_kind":"pith_short_16","alias_value":"DT5KZ6ZIJPQS2D6H","created_at":"2026-06-04T01:08:31.168681+00:00"},{"alias_kind":"pith_short_8","alias_value":"DT5KZ6ZI","created_at":"2026-06-04T01:08:31.168681+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2602.06846","citing_title":"DynFOA: Generating First-Order Ambisonics with Conditional Diffusion for Dynamic and Acoustically Complex 360-Degree Videos","ref_index":25,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DT5KZ6ZIJPQS2D6HL5I4RKGEE4","json":"https://pith.science/pith/DT5KZ6ZIJPQS2D6HL5I4RKGEE4.json","graph_json":"https://pith.science/api/pith-number/DT5KZ6ZIJPQS2D6HL5I4RKGEE4/graph.json","events_json":"https://pith.science/api/pith-number/DT5KZ6ZIJPQS2D6HL5I4RKGEE4/events.json","paper":"https://pith.science/paper/DT5KZ6ZI"},"agent_actions":{"view_html":"https://pith.science/pith/DT5KZ6ZIJPQS2D6HL5I4RKGEE4","download_json":"https://pith.science/pith/DT5KZ6ZIJPQS2D6HL5I4RKGEE4.json","view_paper":"https://pith.science/paper/DT5KZ6ZI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.14623&json=true","fetch_graph":"https://pith.science/api/pith-number/DT5KZ6ZIJPQS2D6HL5I4RKGEE4/graph.json","fetch_events":"https://pith.science/api/pith-number/DT5KZ6ZIJPQS2D6HL5I4RKGEE4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DT5KZ6ZIJPQS2D6HL5I4RKGEE4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DT5KZ6ZIJPQS2D6HL5I4RKGEE4/action/storage_attestation","attest_author":"https://pith.science/pith/DT5KZ6ZIJPQS2D6HL5I4RKGEE4/action/author_attestation","sign_citation":"https://pith.science/pith/DT5KZ6ZIJPQS2D6HL5I4RKGEE4/action/citation_signature","submit_replication":"https://pith.science/pith/DT5KZ6ZIJPQS2D6HL5I4RKGEE4/action/replication_record"}},"created_at":"2026-06-04T01:08:31.168681+00:00","updated_at":"2026-06-04T01:08:31.168681+00:00"}