{"id":"W4372260241","doi":"10.1109/icassp49357.2023.10097115","title":"SARdBScene: Dataset and Resnet Baseline for Audio Scene Source Counting and Analysis","year":2023,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Baseline (sea); Computer science; Task (project management); Audio analyzer; Audio signal processing; Speech recognition; Artificial intelligence; Audio signal; Speech coding; Engineering","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":[0.000909643,0.000095342,0.0001766344,0.0001753952,0.0002368707,0.0002997927,0.0002132188,0.00003255202,0.0000133275],"category_scores_gemma":[0.00009796191,0.00008055388,0.00002684632,0.000908334,0.00003851064,0.0002908342,0.0002877453,0.00004537591,0.00000648146],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000005443754,"about_ca_system_score_gemma":0.00002375166,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00004252392,"about_ca_topic_score_gemma":0.00003092227,"domain_scores_codex":[0.9990297,0.00002331524,0.0001633446,0.0004076232,0.0001426902,0.0002333305],"domain_scores_gemma":[0.9993434,0.0001956143,0.00006265572,0.0002781943,0.00004350981,0.00007664978],"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":[0.00001555343,0.00002797152,0.01515219,0.0002321146,0.0002739963,0.00001373286,0.0006014191,0.0007209204,0.001155503,0.001855095,0.3993503,0.5806012],"study_design_scores_gemma":[0.0002741717,0.00001420194,0.003113403,0.00001404267,0.00008026928,0.000003767223,0.00004016923,0.8751554,0.000517912,0.00028583,0.1203457,0.0001550943],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01584989,0.0001789289,0.9806724,0.002744591,0.00004054354,0.00008274888,0.0001386717,0.0001613347,0.0001308695],"genre_scores_gemma":[0.5759035,0.0003029686,0.4050072,0.01144458,0.0004309811,0.00003765626,0.002382557,0.00003732035,0.004453291],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8744345,"threshold_uncertainty_score":0.3284892,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02702748367579776,"score_gpt":0.2825128516751092,"score_spread":0.2554853679993114,"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."}}