{"id":"W4409796512","doi":"10.1109/southeastcon56624.2025.10971477","title":"Deep Multimodal Classification of Musical Genres","year":2025,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Moncton","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Musical; Computer science; Artificial intelligence; Natural language processing; Speech recognition; Art; Literature","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006119685,0.001007439,0.0006456095,0.0012948,0.0002772704,0.0009320576,0.0005563921,0.0005210271,0.005613384],"category_scores_gemma":[0.001500077,0.0001443018,0.0006879485,0.0007617787,0.0002168655,0.001344157,0.001056979,0.0009148263,0.002382615],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004900603,"about_ca_system_score_gemma":0.0003149126,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002640813,"about_ca_topic_score_gemma":0.004577308,"domain_scores_codex":[0.9996768,0.00006339308,0.00001599868,0.00009311028,0.00007785207,0.00007288056],"domain_scores_gemma":[0.9995891,0.0001164382,0.00003634059,0.00007248546,0.0001464617,0.00003915537],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0003637901,0.0002264693,0.00507491,0.0001629313,0.0001029476,0.0001148617,0.0001462702,0.01735602,0.05905659,0.001948432,0.007113805,0.9083329],"study_design_scores_gemma":[0.0000231643,0.0002411045,0.01097604,0.00008154984,0.0001026194,0.0001513905,0.0004603106,0.9338118,0.03508485,0.008186677,0.01084096,0.00003960158],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.327755,0.004084992,0.6310178,0.001081274,0.0005372704,0.0002150791,0.003352461,0.007676347,0.02427974],"genre_scores_gemma":[0.8732279,0.0008653518,0.1043636,0.0002350009,0.0002058511,0.00008972248,0.00394836,0.0001712994,0.01689295],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005613384,"threshold_uncertainty_score":0.01877862,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02001921047696685,"score_gpt":0.2770585678062852,"score_spread":0.2570393573293183,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}