{"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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001569156,0.0001850144,0.0001638228,0.000124326,0.0001398073,0.0009092857,0.000581555,0.00007228862,0.0001113692],"category_scores_gemma":[0.00002679457,0.0001823035,0.00009329238,0.0003281535,0.0000446437,0.001110003,0.0001908935,0.0002439706,0.0004882647],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00006475871,"about_ca_system_score_gemma":0.00003474635,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001434801,"about_ca_topic_score_gemma":0.000002687925,"domain_scores_codex":[0.9984428,0.0000444044,0.0002060901,0.0006260592,0.0003421619,0.0003385331],"domain_scores_gemma":[0.9991919,0.0002292292,0.00005950894,0.0004050087,0.0000316089,0.00008276558],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.00004178432,0.00003349266,0.0001779412,0.0001542968,0.0001154267,0.0004915591,0.007191314,0.00004637589,0.1600956,0.000153309,0.3280209,0.5034781],"study_design_scores_gemma":[0.002027242,0.00009333696,0.001119989,0.002707832,0.0001363383,0.00003617486,0.000234994,0.1011054,0.3341255,0.004170684,0.5523198,0.001922706],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5538824,0.0008361344,0.4167732,0.02043026,0.003657549,0.0001388601,0.00002976749,0.0007323279,0.003519477],"genre_scores_gemma":[0.8167785,0.00006464725,0.1280951,0.05073256,0.003291192,0.00003766327,0.00006462471,0.0000917017,0.0008440219],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.5015553,"threshold_uncertainty_score":0.8768263,"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."}}