{"id":"W4399778321","doi":"10.1121/10.0026385","title":"Who is singing? Voice recognition from spoken versus sung speech","year":2024,"lang":"en","type":"article","venue":"JASA Express Letters","topic":"Music and Audio Processing","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"Social Sciences and Humanities Research Council","keywords":"Singing; Modality (human–computer interaction); Speech recognition; Modalities; Identity (music); Masking (illustration); Psychology; Computer science; Acoustics; Artificial intelligence; Art","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":[0.001256623,0.0001281356,0.0003381054,0.0001845361,0.0001950652,0.0009303307,0.0001880646,0.000507268,0.003291307],"category_scores_gemma":[0.004698575,0.0001492821,0.0001949166,0.00008618743,0.0004724338,0.001126587,0.000364021,0.0003790328,0.001189477],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000821943,"about_ca_system_score_gemma":0.0001244602,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004359648,"about_ca_topic_score_gemma":0.001118522,"domain_scores_codex":[0.999611,0.0001114291,0.00003571698,0.0001167912,0.00007043449,0.0000547007],"domain_scores_gemma":[0.9986905,0.0006667696,0.000186197,0.0001254359,0.00016562,0.0001654088],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.002429836,0.0001450901,0.1991076,0.0004014017,0.0002069923,0.0006048352,0.005734197,0.0003934024,0.5112468,0.000371431,0.0008785717,0.27848],"study_design_scores_gemma":[0.00005276497,0.00206152,0.8747644,0.0001751548,0.0002282023,0.003307766,0.01090518,0.006244154,0.09666499,0.002653813,0.002850502,0.00009165561],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9960151,0.0003638644,0.00134766,0.0001586174,0.00003583223,0.000007171587,0.00004686651,0.00002052609,0.002004283],"genre_scores_gemma":[0.9984145,0.0002117194,0.000667911,0.00007548964,0.0000186452,0.000004267339,0.0000524886,0.00001022654,0.0005446328],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.003291307,"threshold_uncertainty_score":0.01101053,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0274344952482056,"score_gpt":0.253536801175287,"score_spread":0.2261023059270814,"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."}}