{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:THKRLJLPL5TSG53TKDM62B7T35","short_pith_number":"pith:THKRLJLP","schema_version":"1.0","canonical_sha256":"99d515a56f5f6723777350d9ed07f3df67d793a825f1ef54d91b425104603022","source":{"kind":"arxiv","id":"2603.09714","version":2},"attestation_state":"computed","paper":{"title":"MUGEN: Evaluating and Improving Multi-audio Understanding of Large Audio-Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","eess.AS"],"primary_cat":"cs.SD","authors_text":"Chih-Kai Yang, Hung-Wei Chen, Hung-yi Lee, Ke-Han Lu, Ping-Le Tsai, Ting-Lin Hsiao, Yen-Ting Piao, Yu-Kai Guo, Yun-Man Hsu, Yun-Shao Tsai","submitted_at":"2026-03-10T14:22:22Z","abstract_excerpt":"While multi-audio understanding is critical for large audio-language models (LALMs), it remains underexplored. We introduce MUGEN, a comprehensive benchmark evaluating this capability across speech, general audio, and music. Our experiments reveal consistent weaknesses in multi-audio settings, and performance degrades sharply as the number of concurrent audio inputs increases, identifying input scaling as a fundamental bottleneck. We further investigate training-free strategies and observe that Audio-Permutational Self-Consistency, which diversifies the order of audio candidates, helps models "},"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":"2603.09714","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.SD","submitted_at":"2026-03-10T14:22:22Z","cross_cats_sorted":["cs.AI","cs.CL","eess.AS"],"title_canon_sha256":"6ca8dd7866992e205930ee3399180842c228dd57980f7f7e56939ad7bbeea4d2","abstract_canon_sha256":"d87a40dfafe5a6e4c573b12d7660812c56d171b754979315e42df52aa50342e2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-14T01:20:55.867575Z","signature_b64":"uMofgQiF0AVjhUDvss00e/X9fbyCOywl7rBZ6knVf7cEmzTuIpOsqxZWsJnNuhGeWuv18vi3YBt8JxX4UY0/Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"99d515a56f5f6723777350d9ed07f3df67d793a825f1ef54d91b425104603022","last_reissued_at":"2026-07-14T01:20:55.866619Z","signature_status":"signed_v1","first_computed_at":"2026-07-14T01:20:55.866619Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MUGEN: Evaluating and Improving Multi-audio Understanding of Large Audio-Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","eess.AS"],"primary_cat":"cs.SD","authors_text":"Chih-Kai Yang, Hung-Wei Chen, Hung-yi Lee, Ke-Han Lu, Ping-Le Tsai, Ting-Lin Hsiao, Yen-Ting Piao, Yu-Kai Guo, Yun-Man Hsu, Yun-Shao Tsai","submitted_at":"2026-03-10T14:22:22Z","abstract_excerpt":"While multi-audio understanding is critical for large audio-language models (LALMs), it remains underexplored. We introduce MUGEN, a comprehensive benchmark evaluating this capability across speech, general audio, and music. Our experiments reveal consistent weaknesses in multi-audio settings, and performance degrades sharply as the number of concurrent audio inputs increases, identifying input scaling as a fundamental bottleneck. We further investigate training-free strategies and observe that Audio-Permutational Self-Consistency, which diversifies the order of audio candidates, helps models "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2603.09714","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/2603.09714/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":"2603.09714","created_at":"2026-07-14T01:20:55.867035+00:00"},{"alias_kind":"arxiv_version","alias_value":"2603.09714v2","created_at":"2026-07-14T01:20:55.867035+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2603.09714","created_at":"2026-07-14T01:20:55.867035+00:00"},{"alias_kind":"pith_short_12","alias_value":"THKRLJLPL5TS","created_at":"2026-07-14T01:20:55.867035+00:00"},{"alias_kind":"pith_short_16","alias_value":"THKRLJLPL5TSG53T","created_at":"2026-07-14T01:20:55.867035+00:00"},{"alias_kind":"pith_short_8","alias_value":"THKRLJLP","created_at":"2026-07-14T01:20:55.867035+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2604.24401","citing_title":"All That Glitters Is Not Audio: Rethinking Text Priors and Audio Reliance in Audio-Language Evaluation","ref_index":40,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/THKRLJLPL5TSG53TKDM62B7T35","json":"https://pith.science/pith/THKRLJLPL5TSG53TKDM62B7T35.json","graph_json":"https://pith.science/api/pith-number/THKRLJLPL5TSG53TKDM62B7T35/graph.json","events_json":"https://pith.science/api/pith-number/THKRLJLPL5TSG53TKDM62B7T35/events.json","paper":"https://pith.science/paper/THKRLJLP"},"agent_actions":{"view_html":"https://pith.science/pith/THKRLJLPL5TSG53TKDM62B7T35","download_json":"https://pith.science/pith/THKRLJLPL5TSG53TKDM62B7T35.json","view_paper":"https://pith.science/paper/THKRLJLP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2603.09714&json=true","fetch_graph":"https://pith.science/api/pith-number/THKRLJLPL5TSG53TKDM62B7T35/graph.json","fetch_events":"https://pith.science/api/pith-number/THKRLJLPL5TSG53TKDM62B7T35/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/THKRLJLPL5TSG53TKDM62B7T35/action/timestamp_anchor","attest_storage":"https://pith.science/pith/THKRLJLPL5TSG53TKDM62B7T35/action/storage_attestation","attest_author":"https://pith.science/pith/THKRLJLPL5TSG53TKDM62B7T35/action/author_attestation","sign_citation":"https://pith.science/pith/THKRLJLPL5TSG53TKDM62B7T35/action/citation_signature","submit_replication":"https://pith.science/pith/THKRLJLPL5TSG53TKDM62B7T35/action/replication_record"}},"created_at":"2026-07-14T01:20:55.867035+00:00","updated_at":"2026-07-14T01:20:55.867035+00:00"}