{"id":"W4386246038","doi":"10.32920/24050751","title":"Multimodal System for Audio Scene Source Counting and Analysis","year":2023,"lang":"en","type":"preprint","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Computer science; Speech recognition; Margin (machine learning); Set (abstract data type); Sound recording and reproduction; Audio signal processing; Audio analyzer; Event (particle physics); Task (project management); Modality (human–computer interaction); Artificial intelligence; Pattern recognition (psychology); Audio signal; Speech coding; Machine learning","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007171113,0.0002302699,0.0005020222,0.0003047255,0.0002557406,0.0007942287,0.0007154914,0.0001752378,0.000003028939],"category_scores_gemma":[0.0000494725,0.0002040281,0.0002088864,0.0004989642,0.00003224134,0.0001385644,0.00158832,0.0001961595,0.00001109199],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006278578,"about_ca_system_score_gemma":0.0001063374,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002437934,"about_ca_topic_score_gemma":0.00003021044,"domain_scores_codex":[0.9981562,0.00002954564,0.0003470402,0.0008841944,0.0002643336,0.0003186093],"domain_scores_gemma":[0.9987382,0.0001685929,0.0002815583,0.0005702431,0.000155729,0.00008571627],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00003728515,0.0001471487,0.04126679,0.02049611,0.007176597,0.00007192777,0.009180544,0.1556193,0.001011043,0.02360778,0.01350538,0.7278801],"study_design_scores_gemma":[0.0001525291,0.000005492079,0.001301739,0.0001918533,0.0002218079,0.000002404731,0.0000862981,0.9966284,0.0002829148,0.0003032728,0.0005550763,0.0002682045],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.007128564,0.0001535008,0.9899421,0.0006504391,0.0004863977,0.0002508749,0.000009619498,0.0007496572,0.0006287908],"genre_scores_gemma":[0.6842729,0.00000863137,0.3118886,0.0002435652,0.0002694267,0.0000771848,0.00002489406,0.00002614197,0.003188659],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8410091,"threshold_uncertainty_score":0.8320026,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03470614940164925,"score_gpt":0.2715294978241795,"score_spread":0.2368233484225302,"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."}}