{"id":"W4386256511","doi":"10.32920/24050751.v1","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":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Computer science; Speech recognition; Margin (machine learning); Set (abstract data type); Audio analyzer; Audio signal processing; Sound recording and reproduction; 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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007091684,0.001556558,0.001016201,0.001520079,0.0004064036,0.00109982,0.001686336,0.001275907,0.04181952],"category_scores_gemma":[0.001469444,0.0003796378,0.0007820029,0.0007767403,0.0002655815,0.001700295,0.00240879,0.0009820014,0.03094322],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006450723,"about_ca_system_score_gemma":0.0006415561,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004112397,"about_ca_topic_score_gemma":0.006630454,"domain_scores_codex":[0.999454,0.00006341353,0.00003010874,0.0002110233,0.0001680222,0.00007346032],"domain_scores_gemma":[0.9995922,0.00004211593,0.00002386197,0.0001307597,0.000173214,0.0000377555],"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":[0.001364121,0.000214694,0.001130847,0.0005345221,0.0002030266,0.0003440876,0.000133402,0.003998161,0.1081295,0.004080546,0.1770713,0.7027959],"study_design_scores_gemma":[0.0003675994,0.000785241,0.01217423,0.0002535276,0.000285471,0.001241249,0.0002982492,0.4740748,0.1880091,0.0192913,0.3029384,0.0002809543],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02965342,0.002870903,0.6305922,0.0006849998,0.001262342,0.001057863,0.05028187,0.2474437,0.03615274],"genre_scores_gemma":[0.2347234,0.001290644,0.6065037,0.001314631,0.0005586926,0.001824712,0.09287988,0.006022,0.0548824],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.04181952,"threshold_uncertainty_score":0.1399002,"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."}}