{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:TYNRTIPQHLSCW7MVJCWERHQVGY","short_pith_number":"pith:TYNRTIPQ","schema_version":"1.0","canonical_sha256":"9e1b19a1f03ae42b7d9548ac489e15362e2d0d88d52d2011ae1bed8e54eda4eb","source":{"kind":"arxiv","id":"2605.17225","version":1},"attestation_state":"computed","paper":{"title":"Can Large Audio Language Models Ignore Multilingual Distractors? An Evaluation of Their Selective Auditory Attention Capabilities","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"Large audio language models lose selective attention to English targets when multilingual distractors appear at low signal-to-noise ratios.","cross_cats":[],"primary_cat":"eess.AS","authors_text":"Heejoon Koo","submitted_at":"2026-05-17T02:13:58Z","abstract_excerpt":"Robust selective auditory attention under multilingual interference is critical for reliable deployment of Large Audio Language Models (LALMs). We introduce MUSA, a cocktail party-inspired multilingual benchmark for source-grounded spoken-language understanding and reasoning. Each item pairs an English target dialogue with a semantically plausible distractor in English, Spanish, Korean, or Chinese, and evaluates models across (1) single, (2) source separation-based two-stage, (3) and end-to-end cocktail party settings under controlled SNRs. Evaluating two closed-source and four open-weight LAL"},"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":true,"formal_links_present":false},"canonical_record":{"source":{"id":"2605.17225","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"eess.AS","submitted_at":"2026-05-17T02:13:58Z","cross_cats_sorted":[],"title_canon_sha256":"91eca395424920448f8f8f0f2ed0fd400187c916b1602de2e5dc94304d7094b6","abstract_canon_sha256":"d2cee1fd02cd40e01e50d8fb6b2439be51edf9592182a086912856951b922371"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T00:03:46.148167Z","signature_b64":"e5CKXBSCduyAXFeZ1HC8naUTg4iGQkc2ctphzJM6UAch3s5eL6A1KmQ43wN1a2FjPlyIWR0OaQ9g905ri1WqAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9e1b19a1f03ae42b7d9548ac489e15362e2d0d88d52d2011ae1bed8e54eda4eb","last_reissued_at":"2026-05-20T00:03:46.147257Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T00:03:46.147257Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Can Large Audio Language Models Ignore Multilingual Distractors? An Evaluation of Their Selective Auditory Attention Capabilities","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"Large audio language models lose selective attention to English targets when multilingual distractors appear at low signal-to-noise ratios.","cross_cats":[],"primary_cat":"eess.AS","authors_text":"Heejoon Koo","submitted_at":"2026-05-17T02:13:58Z","abstract_excerpt":"Robust selective auditory attention under multilingual interference is critical for reliable deployment of Large Audio Language Models (LALMs). We introduce MUSA, a cocktail party-inspired multilingual benchmark for source-grounded spoken-language understanding and reasoning. Each item pairs an English target dialogue with a semantically plausible distractor in English, Spanish, Korean, or Chinese, and evaluates models across (1) single, (2) source separation-based two-stage, (3) and end-to-end cocktail party settings under controlled SNRs. Evaluating two closed-source and four open-weight LAL"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"strong single performance does not ensure robust selective auditory attention: cocktail party accuracy degrades under severe SNRs, and errors are dominated by distractor-grounded source confusion. In addition, separation reduces acoustic overlap but leaves source attribution unresolved, often yielding confident wrong-stream answers.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The MUSA benchmark pairs English targets with semantically plausible distractors under controlled SNRs in single, two-stage, and end-to-end settings, assuming this construction accurately measures selective auditory attention without major confounding factors from the specific dialogue content or model training data.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Introduces the MUSA benchmark and evaluates LALMs showing that strong single-speaker performance fails to ensure robust selective attention under multilingual interference, with errors from source confusion and unresolved attribution after separation.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Large audio language models lose selective attention to English targets when multilingual distractors appear at low signal-to-noise ratios.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"af4eecb5ae7658a758cf42181ba7f163a1650b29dec36763b7b2f426c16d3c39"},"source":{"id":"2605.17225","kind":"arxiv","version":1},"verdict":{"id":"b5bfd01b-fc3a-455d-a9a9-bb7da9626d62","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T23:13:12.660168Z","strongest_claim":"strong single performance does not ensure robust selective auditory attention: cocktail party accuracy degrades under severe SNRs, and errors are dominated by distractor-grounded source confusion. In addition, separation reduces acoustic overlap but leaves source attribution unresolved, often yielding confident wrong-stream answers.","one_line_summary":"Introduces the MUSA benchmark and evaluates LALMs showing that strong single-speaker performance fails to ensure robust selective attention under multilingual interference, with errors from source confusion and unresolved attribution after separation.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The MUSA benchmark pairs English targets with semantically plausible distractors under controlled SNRs in single, two-stage, and end-to-end settings, assuming this construction accurately measures selective auditory attention without major confounding factors from the specific dialogue content or model training data.","pith_extraction_headline":"Large audio language models lose selective attention to English targets when multilingual distractors appear at low signal-to-noise ratios."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.17225/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T23:31:20.355347Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T23:22:22.898125Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T22:01:57.910606Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T21:33:23.806626Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"d10e32fc2de6b09d0c700f4c69f49ad235b1fb3e55dc342da97bcfa70b99052f"},"references":{"count":36,"sample":[{"doi":"","year":2017,"title":"Situational awareness , pages=","work_id":"7199963f-dbe2-4eea-a620-4532164d80f6","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Journal of the acoustical society of America , volume=","work_id":"6516012c-659c-4ffe-9b11-610c6f874bad","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=","work_id":"2e9aa9ac-c395-45a3-b980-49925a891304","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"arXiv preprint arXiv:2501.04962 , year=","work_id":"674808a4-e636-424c-882e-72318e2120af","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Audiotrust: Benchmarking the multifaceted trustworthiness of audio large language models","work_id":"ddc276b0-adb9-4470-99e0-fa7f1b36a213","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":36,"snapshot_sha256":"93a9fbadc82b4f2e6e2382834a84ec6a37544ea14d22437d901222ee40bd0ba3","internal_anchors":7},"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":"2605.17225","created_at":"2026-05-20T00:03:46.147414+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.17225v1","created_at":"2026-05-20T00:03:46.147414+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.17225","created_at":"2026-05-20T00:03:46.147414+00:00"},{"alias_kind":"pith_short_12","alias_value":"TYNRTIPQHLSC","created_at":"2026-05-20T00:03:46.147414+00:00"},{"alias_kind":"pith_short_16","alias_value":"TYNRTIPQHLSCW7MV","created_at":"2026-05-20T00:03:46.147414+00:00"},{"alias_kind":"pith_short_8","alias_value":"TYNRTIPQ","created_at":"2026-05-20T00:03:46.147414+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/TYNRTIPQHLSCW7MVJCWERHQVGY","json":"https://pith.science/pith/TYNRTIPQHLSCW7MVJCWERHQVGY.json","graph_json":"https://pith.science/api/pith-number/TYNRTIPQHLSCW7MVJCWERHQVGY/graph.json","events_json":"https://pith.science/api/pith-number/TYNRTIPQHLSCW7MVJCWERHQVGY/events.json","paper":"https://pith.science/paper/TYNRTIPQ"},"agent_actions":{"view_html":"https://pith.science/pith/TYNRTIPQHLSCW7MVJCWERHQVGY","download_json":"https://pith.science/pith/TYNRTIPQHLSCW7MVJCWERHQVGY.json","view_paper":"https://pith.science/paper/TYNRTIPQ","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.17225&json=true","fetch_graph":"https://pith.science/api/pith-number/TYNRTIPQHLSCW7MVJCWERHQVGY/graph.json","fetch_events":"https://pith.science/api/pith-number/TYNRTIPQHLSCW7MVJCWERHQVGY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TYNRTIPQHLSCW7MVJCWERHQVGY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TYNRTIPQHLSCW7MVJCWERHQVGY/action/storage_attestation","attest_author":"https://pith.science/pith/TYNRTIPQHLSCW7MVJCWERHQVGY/action/author_attestation","sign_citation":"https://pith.science/pith/TYNRTIPQHLSCW7MVJCWERHQVGY/action/citation_signature","submit_replication":"https://pith.science/pith/TYNRTIPQHLSCW7MVJCWERHQVGY/action/replication_record"}},"created_at":"2026-05-20T00:03:46.147414+00:00","updated_at":"2026-05-20T00:03:46.147414+00:00"}