{"meta":{"page":1,"per_page":50,"max_per_page":100,"total":1031,"total_is_capped":false,"direct_labels_cover":2,"predictions_cover":1031,"direct_label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline (scores rank; they never assert a category)","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12","author_layer_release":"2026-06-26"},"query_hash":"7a36b2ebf8e4","filters":{"topic":"Music and Audio Processing"}},"results":[{"id":"W2099866409","doi":"10.1145/1273496.1273596","title":"Restricted Boltzmann machines for collaborative filtering","year":2007,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":1882,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Restricted Boltzmann machine; Boltzmann machine; Computer science; Collaborative filtering; Graphical model; Inference; Set (abstract data type); Class (philosophy); Artificial intelligence; Data set; Machine learning; Layer (electronics); Recommender system; Data mining; Deep learning","authors":[{"name":"Ruslan Salakhutdinov","is_ca":true},{"name":"Andriy Mnih","is_ca":true},{"name":"Geoffrey E. Hinton","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02414147631483186,"gpt":0.2826722957050696,"spread":0.2585308193902377,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004304946,0.001412944,0.003124688,0.001098505,0.0008672411,0.001931543,0.003627263,0.0029681,0.004459522],"category_scores_gemma":[0.01911542,0.001201264,0.001996533,0.001983974,0.00200367,0.003438915,0.002441729,0.004180862,0.001992662],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001852401,"about_ca_system_score_gemma":0.001653415,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009789172,"about_ca_topic_score_gemma":0.006741953,"domain_scores_codex":[0.9962154,0.001815019,0.0001907639,0.0007606056,0.0007122729,0.0003058707],"domain_scores_gemma":[0.9934953,0.004746336,0.0003176705,0.0008082286,0.0004780438,0.0001544542],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00009459788,0.00005902309,0.0007074632,0.0001420099,0.0001751328,0.0000604907,0.0001026886,0.7986952,0.0006115727,0.1281053,0.003026125,0.06822044],"study_design_scores_gemma":[0.000008292218,0.00000911086,0.00006026435,0.000007871974,0.000007687766,0.00001294438,0.000004414935,0.9318132,0.0001511092,0.06725008,0.0006651763,0.000009918402],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.002055134,0.0004096077,0.9957903,0.0002833849,0.00005288016,0.00002479152,0.00007284418,0.0003906089,0.0009204106],"genre_scores_gemma":[0.4254025,0.001999518,0.5581459,0.000918903,0.0005271389,0.0008994011,0.001075479,0.0004051997,0.01062598],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009789172,"threshold_uncertainty_score":0.02276701,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2053101950","doi":"10.1007/s00530-010-0182-0","title":"Multimodal fusion for multimedia analysis: a survey","year":2010,"lang":"en","type":"article","venue":"Multimedia Systems","topic":"Music and Audio Processing","field":"Computer Science","cited_by":1226,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa; University of Winnipeg","funders":"","keywords":"Computer science; Modalities; Modality (human–computer interaction); Perspective (graphical); Process (computing); Multimedia; Sensor fusion; Artificial intelligence; Human–computer interaction","authors":[{"name":"Pradeep K. Atrey","is_ca":true},{"name":"M. Anwar Hossain","is_ca":true},{"name":"Abdulmotaleb El Saddik","is_ca":true},{"name":"Mohan Kankanhalli","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02559643849957117,"gpt":0.2733577503753194,"spread":0.2477613118757482,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00199401,0.001028727,0.00179577,0.002777794,0.0004089171,0.002234374,0.001660667,0.001336041,0.003771662],"category_scores_gemma":[0.002944856,0.0005167444,0.001067796,0.004458923,0.0004700835,0.002143037,0.001035275,0.000770207,0.002290418],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002925216,"about_ca_system_score_gemma":0.0005376106,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001171543,"about_ca_topic_score_gemma":0.001271761,"domain_scores_codex":[0.9989875,0.0002023689,0.00009352479,0.0002265511,0.0004440521,0.00004605623],"domain_scores_gemma":[0.9981446,0.00103737,0.00008948555,0.0001567091,0.0005229224,0.00004883895],"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":[0.0001059091,0.0001077522,0.001175317,0.001597109,0.0001468826,0.00008996952,0.00009335991,0.003879466,0.008628327,0.004166834,0.003698799,0.9763103],"study_design_scores_gemma":[0.000151999,0.001486564,0.01605362,0.002195655,0.001238363,0.005997206,0.001516655,0.4692806,0.08634052,0.06159499,0.3536768,0.0004670658],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"review","genre_scores_codex":[0.01128457,0.2460981,0.7292264,0.0006895884,0.0003529435,0.0001632509,0.0003277049,0.0009354521,0.01092202],"genre_scores_gemma":[0.1458811,0.3971932,0.4400711,0.0006427385,0.002635612,0.0003637983,0.001478289,0.0003868162,0.01134742],"genre_candidate":"review","genre_consensus":null,"teacher_disagreement_score":0.003771662,"threshold_uncertainty_score":0.01261747,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1718516643","doi":"10.1073/pnas.1414495112","title":"Statistical universals reveal the structures and functions of human music","year":2015,"lang":"en","type":"article","venue":"Proceedings of the National Academy of Sciences","topic":"Music and Audio Processing","field":"Computer Science","cited_by":650,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McMaster University","funders":"","keywords":"Problem of universals; Linguistics; Psychology; Mathematics; Philosophy","authors":[{"name":"Patrick E. Savage","is_ca":false},{"name":"Steven Brown","is_ca":true},{"name":"Emi Sakai","is_ca":false},{"name":"Thomas E. Currie","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.08503047264000195,"gpt":0.3169157324849309,"spread":0.2318852598449289,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002697126,0.0003263493,0.0005402092,0.003281167,0.001335746,0.0018036,0.0007170748,0.0004623395,0.003142694],"category_scores_gemma":[0.01887481,0.0002855016,0.0005877975,0.003155756,0.004323327,0.002455329,0.002372243,0.001085285,0.0002601358],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006402062,"about_ca_system_score_gemma":0.000530577,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002690426,"about_ca_topic_score_gemma":0.003144919,"domain_scores_codex":[0.998021,0.0005748649,0.0001412806,0.0008109814,0.0003167804,0.00013506],"domain_scores_gemma":[0.9863119,0.007401586,0.00218225,0.002840867,0.0008920978,0.0003713724],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0003296833,0.00006892701,0.6779838,0.0005807378,0.0008531218,0.0004325554,0.008322772,0.006872111,0.03124822,0.05921393,0.001061709,0.2130325],"study_design_scores_gemma":[0.00001506963,0.0001556949,0.8727964,0.00006516302,0.0002217892,0.0006991342,0.003919488,0.01572545,0.002859385,0.09871296,0.004751089,0.00007821128],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9272411,0.001070675,0.06317445,0.0004252342,0.00002446394,0.00003071236,0.0006275705,0.0002422848,0.007163482],"genre_scores_gemma":[0.9953883,0.0001046053,0.004094452,0.00003134748,0.00001435722,0.00001191521,0.0002037019,0.0000246593,0.000126556],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.003281167,"threshold_uncertainty_score":0.01426393,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2964335273","doi":"","title":"How to Construct Deep Recurrent Neural Networks","year":2014,"lang":"en","type":"article","venue":"International Conference on Learning Representations","topic":"Music and Audio Processing","field":"Computer Science","cited_by":582,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal","funders":"","keywords":"Recurrent neural network; Computer science; Deep learning; Artificial intelligence; Construct (python library); Feedforward neural network; Feed forward; Function (biology); Artificial neural network; Engineering","authors":[{"name":"Razvan Pascanu","is_ca":true},{"name":"Çağlar Gülçehre","is_ca":true},{"name":"Kyunghyun Cho","is_ca":false},{"name":"Yoshua Bengio","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03925941227028947,"gpt":0.3147329136198822,"spread":0.2754735013495928,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009555257,0.0007502528,0.0005308024,0.0004996305,0.0002797953,0.0008316292,0.0009771748,0.000855897,0.003522978],"category_scores_gemma":[0.003507933,0.0004668619,0.0006068601,0.0003782953,0.0004477171,0.002041165,0.001095475,0.001319597,0.001234269],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006235508,"about_ca_system_score_gemma":0.0006437092,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002513854,"about_ca_topic_score_gemma":0.003824462,"domain_scores_codex":[0.9996511,0.00009173994,0.00002959805,0.00008572071,0.00009134743,0.00005053],"domain_scores_gemma":[0.999294,0.0002845119,0.00007163855,0.0001069882,0.0002038544,0.000039038],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001526227,0.00008483396,0.001986124,0.0002485861,0.0001356953,0.0002578394,0.0001821272,0.5345149,0.01976183,0.0723472,0.007049818,0.3632784],"study_design_scores_gemma":[0.000006741291,0.00002388981,0.0001026934,0.00002262358,0.00001340801,0.0000251038,0.00001319272,0.9756703,0.003336464,0.01892612,0.001852758,0.000006822947],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01153755,0.0002303031,0.9848915,0.0002966361,0.0000651156,0.00002951292,0.00007707856,0.0009579756,0.001914173],"genre_scores_gemma":[0.3857969,0.0004550605,0.6073133,0.0003603858,0.00007968955,0.0001317726,0.0004112747,0.0003028155,0.005148723],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003522978,"threshold_uncertainty_score":0.01178557,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1819710477","doi":"10.48550/arxiv.1206.6392","title":"Modeling Temporal Dependencies in High-Dimensional Sequences: Application to Polyphonic Music Generation and Transcription","year":2012,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Music and Audio Processing","field":"Computer Science","cited_by":490,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal","funders":"","keywords":"Polyphony; Computer science; Estimator; Probabilistic logic; Speech recognition; Representation (politics); Transcription (linguistics); Artificial intelligence; Statistical model; Natural language processing; Mathematics; Linguistics","authors":[{"name":"Nicolas Boulanger-Lewandowski","is_ca":true},{"name":"Yoshua Bengio","is_ca":true},{"name":"Pascal Vincent","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.09894440991759594,"gpt":0.1934819880977269,"spread":0.09453757818013096,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00126,0.0006203183,0.0006292033,0.0006575224,0.0003860176,0.0007379783,0.0009735868,0.001091489,0.001276375],"category_scores_gemma":[0.007726796,0.0003956055,0.0005671773,0.0009503207,0.0005425877,0.001148786,0.0006806192,0.001373841,0.0004212943],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005490187,"about_ca_system_score_gemma":0.0006610476,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005286286,"about_ca_topic_score_gemma":0.005449756,"domain_scores_codex":[0.9995199,0.0001635716,0.00002529443,0.0001545794,0.00009434175,0.00004225925],"domain_scores_gemma":[0.9970124,0.002185511,0.000304347,0.0002423094,0.0001621892,0.00009313866],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001675925,0.00007911702,0.002839269,0.00008167285,0.00004943237,0.0002367289,0.0001466353,0.8556705,0.01098544,0.007046366,0.0007251506,0.1219721],"study_design_scores_gemma":[0.000003920746,0.00001238181,0.0003264498,0.000002293479,0.00000310642,0.00001798027,0.000005222377,0.9962893,0.0007940985,0.002392045,0.0001485937,0.000004751049],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1264305,0.0004925744,0.8707209,0.0004959214,0.00003994541,0.00003682658,0.0003846762,0.0005781394,0.0008205537],"genre_scores_gemma":[0.8475285,0.0005907266,0.1481224,0.0001015037,0.0001200207,0.00008528065,0.001174909,0.0001141225,0.002162506],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005286286,"threshold_uncertainty_score":0.01051104,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2976594877","doi":"10.1016/j.apacoust.2019.107020","title":"Trends in audio signal feature extraction methods","year":2019,"lang":"en","type":"article","venue":"Applied Acoustics","topic":"Music and Audio Processing","field":"Computer Science","cited_by":467,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Toronto Metropolitan University","funders":"Natural Sciences and Engineering Research Council of Canada; Ryerson University","keywords":"Computer science; Feature extraction; SIGNAL (programming language); Audio signal; Speech recognition; Signal processing; Audio signal processing; Frequency domain; Wavelet; Mel-frequency cepstrum; Time domain; Feature (linguistics); Pattern recognition (psychology); Artificial intelligence; Domain (mathematical analysis); Cepstrum; Speech processing; Audio signal flow; Digital signal processing; Speech coding; Computer vision; Mathematics","authors":[{"name":"Garima Sharma","is_ca":true},{"name":"Kartikeyan Umapathy","is_ca":true},{"name":"Sridhar Krishnan","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01591782837905122,"gpt":0.3086868783877196,"spread":0.2927690500086684,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004590762,0.001105921,0.001504419,0.005024853,0.0004488058,0.003067542,0.002104547,0.001542446,0.008930773],"category_scores_gemma":[0.01253328,0.0006713891,0.0008349029,0.005531195,0.001232342,0.005227207,0.001139041,0.002078871,0.005735922],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001024902,"about_ca_system_score_gemma":0.001769622,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001265662,"about_ca_topic_score_gemma":0.001434996,"domain_scores_codex":[0.9970078,0.0004245689,0.000432237,0.0007895378,0.001215074,0.0001308603],"domain_scores_gemma":[0.9807674,0.008001057,0.00106787,0.0009712121,0.008711128,0.0004813048],"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":[0.000209636,0.0001688086,0.001937105,0.002504249,0.00006737193,0.00004128828,0.00008248137,0.001115597,0.01458402,0.007555469,0.009748237,0.9619858],"study_design_scores_gemma":[0.0001671507,0.001559198,0.02841634,0.002920854,0.0005211561,0.002718846,0.001045781,0.08556924,0.08114722,0.04598581,0.7496026,0.0003457708],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"review","genre_scores_codex":[0.02531511,0.4348216,0.4881881,0.01669011,0.00489002,0.0003210312,0.001267048,0.002135207,0.02637175],"genre_scores_gemma":[0.1258528,0.3846842,0.4308644,0.005799724,0.01414526,0.0005033868,0.003338693,0.0008018604,0.03400974],"genre_candidate":"review","genre_consensus":null,"teacher_disagreement_score":0.008930773,"threshold_uncertainty_score":0.02987641,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1964812476","doi":"10.1109/taslp.2014.2303296","title":"Application of Deep Belief Networks for Natural Language Understanding","year":2014,"lang":"en","type":"article","venue":"IEEE/ACM Transactions on Audio Speech and Language Processing","topic":"Music and Audio Processing","field":"Computer Science","cited_by":460,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Deep belief network; Artificial intelligence; Computer science; Support vector machine; Boosting (machine learning); Artificial neural network; Machine learning; Deep learning; Principle of maximum entropy; Backpropagation; Pattern recognition (psychology)","authors":[{"name":"Ruhi Sarikaya","is_ca":false},{"name":"Geoffrey E. Hinton","is_ca":true},{"name":"Anoop Deoras","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01281846356590224,"gpt":0.2571697326409619,"spread":0.2443512690750596,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001467631,0.0008372595,0.0005063832,0.00112688,0.0003403481,0.001424129,0.0008660565,0.001080527,0.002071298],"category_scores_gemma":[0.005941058,0.0004164355,0.0004685527,0.0008515335,0.0006530804,0.002166464,0.001061917,0.001967009,0.0004080087],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001165906,"about_ca_system_score_gemma":0.0008302744,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005178678,"about_ca_topic_score_gemma":0.005768038,"domain_scores_codex":[0.9993916,0.0002646251,0.00003207752,0.0001036293,0.0001642387,0.00004378518],"domain_scores_gemma":[0.9975562,0.001823908,0.0001299791,0.000143322,0.0002952041,0.00005140122],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001207126,0.0001123558,0.001514209,0.0001787109,0.0001139124,0.00009602914,0.0001754312,0.7213684,0.00347487,0.03170686,0.002523675,0.2386149],"study_design_scores_gemma":[0.000003465243,0.000008684933,0.0000965917,0.00001012003,0.000005369392,0.0000083519,0.00001037058,0.9750819,0.0007594979,0.02341662,0.0005949557,0.000004077458],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02473487,0.002049386,0.9656041,0.001562424,0.00008012944,0.00004374071,0.0002117808,0.001372492,0.004341049],"genre_scores_gemma":[0.7466159,0.001430802,0.2483276,0.0003676456,0.0001078905,0.00008036385,0.000411141,0.0001192531,0.002539478],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005178678,"threshold_uncertainty_score":0.01029712,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4372260310","doi":"10.1109/icassp49357.2023.10095969","title":"Large-Scale Contrastive Language-Audio Pretraining with Feature Fusion and Keyword-to-Caption Augmentation","year":2023,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":385,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Mila - Quebec Artificial Intelligence Institute","funders":"","keywords":"Computer science; Audio mining; Natural language processing; Pipeline (software); Speech recognition; Artificial intelligence; Encoder; Construct (python library); Feature (linguistics); Natural language; Language model; Speech processing; Acoustic model; Linguistics","authors":[{"name":"Yusong Wu","is_ca":true},{"name":"Ke Chen","is_ca":false},{"name":"Tianyu Zhang","is_ca":true},{"name":"Yuchen Hui","is_ca":true},{"name":"Taylor Berg-Kirkpatrick","is_ca":false},{"name":"Shlomo Dubnov","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.00899152659473625,"gpt":0.2488691384566329,"spread":0.2398776118618966,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001270851,0.001743469,0.0009132901,0.0008115565,0.0004101219,0.0006636875,0.002643388,0.001437339,0.005466122],"category_scores_gemma":[0.003837546,0.0005868577,0.001310511,0.000704349,0.0007602539,0.002048927,0.001948971,0.002646201,0.002889117],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009679787,"about_ca_system_score_gemma":0.001084727,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005359553,"about_ca_topic_score_gemma":0.008192667,"domain_scores_codex":[0.9992831,0.0001629465,0.00003141254,0.0002862998,0.0001428941,0.00009319477],"domain_scores_gemma":[0.9986265,0.0007540575,0.00007450895,0.000191252,0.0002820877,0.00007158618],"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":[0.0007685081,0.0006944359,0.001629032,0.0002777838,0.0001740924,0.0002570602,0.0001840742,0.1224904,0.06912414,0.002621716,0.01254507,0.7892338],"study_design_scores_gemma":[0.00003490027,0.0002243682,0.0005227529,0.00001760269,0.00003865797,0.0000894573,0.00004240269,0.9641606,0.03067479,0.001871727,0.002298339,0.00002446863],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.05716269,0.0008006398,0.92123,0.0003682797,0.0002041539,0.0002828289,0.0006759436,0.01492501,0.004350474],"genre_scores_gemma":[0.5205704,0.0003439093,0.4600874,0.000974104,0.0001392692,0.0007220655,0.004517694,0.0007113089,0.01193375],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005466122,"threshold_uncertainty_score":0.01828605,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2008066450","doi":"10.1121/1.3642604","title":"The Timbre Toolbox: Extracting audio descriptors from musical signals","year":2011,"lang":"en","type":"article","venue":"The Journal of the Acoustical Society of America","topic":"Music and Audio Processing","field":"Computer Science","cited_by":379,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University; Centre for Interdisciplinary Research in Music Media and Technology","funders":"","keywords":"Timbre; Computer science; Music information retrieval; Speech recognition; Audio signal; Toolbox; Audio analyzer; Pattern recognition (psychology); Set (abstract data type); Audio signal processing; Artificial intelligence; Speech coding; Musical","authors":[{"name":"Geoffroy Peeters","is_ca":false},{"name":"Bruno L. Giordano","is_ca":true},{"name":"Patrick Susini","is_ca":false},{"name":"Nicolas Misdariis","is_ca":false},{"name":"Stephen McAdams","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03284561586498905,"gpt":0.2422844526021924,"spread":0.2094388367372033,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009570363,0.001114569,0.0006760196,0.005584489,0.0004636105,0.00171425,0.0007919744,0.0006784898,0.009338727],"category_scores_gemma":[0.003950893,0.0002897053,0.0008332184,0.002920216,0.0004494039,0.001968254,0.001165051,0.0006071054,0.005942286],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002517402,"about_ca_system_score_gemma":0.0005320774,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001383969,"about_ca_topic_score_gemma":0.001826851,"domain_scores_codex":[0.999576,0.00006391187,0.00003350804,0.00008602841,0.0001965452,0.00004410661],"domain_scores_gemma":[0.9992552,0.0002261135,0.0001414696,0.0001215032,0.0001991933,0.00005641792],"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":[0.0009418328,0.0001558118,0.004151837,0.0007647202,0.0001203911,0.0003409632,0.000364976,0.006594555,0.1180222,0.007274423,0.009868417,0.8514],"study_design_scores_gemma":[0.0004519295,0.001361961,0.1800134,0.0005903831,0.000380046,0.003682043,0.001950781,0.4445446,0.1919653,0.03893398,0.135482,0.0006436275],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"software","genre_scores_codex":[0.06046338,0.001186575,0.9111452,0.0001441948,0.0001266769,0.0004419516,0.006700947,0.009410174,0.01038104],"genre_scores_gemma":[0.2091829,0.001629471,0.766618,0.00009771883,0.0001818866,0.0006831434,0.01062404,0.0008843454,0.0100985],"genre_candidate":"software","genre_consensus":null,"teacher_disagreement_score":0.009338727,"threshold_uncertainty_score":0.03124118,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2126203737","doi":"10.1109/asru.2009.5372931","title":"Unsupervised spoken keyword spotting via segmental DTW on Gaussian posteriorgrams","year":2009,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":338,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Keyword spotting; Computer science; Dynamic time warping; Speech recognition; Artificial intelligence; TIMIT; Ranking (information retrieval); Gaussian; Spotting; Viterbi algorithm; Gaussian process; Natural language processing; Pattern recognition (psychology); Hidden Markov model","authors":[{"name":"Yaodong Zhang","is_ca":true},{"name":"James Glass","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01218157054232683,"gpt":0.2340915820983202,"spread":0.2219100115559933,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007930902,0.000869933,0.0009063541,0.001152565,0.0002704611,0.0005481774,0.0009579262,0.0006482904,0.001581764],"category_scores_gemma":[0.002860628,0.0003096218,0.0005206192,0.001234975,0.0005056833,0.001230209,0.000911046,0.0008420949,0.001410083],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003898407,"about_ca_system_score_gemma":0.0007091792,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003312098,"about_ca_topic_score_gemma":0.004594367,"domain_scores_codex":[0.9993517,0.0001734324,0.00003425038,0.0002155748,0.0001560872,0.00006910531],"domain_scores_gemma":[0.9988548,0.000572736,0.0001275996,0.000130417,0.0002592369,0.00005525097],"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":[0.0006080598,0.0001162296,0.001384608,0.000154055,0.00008709407,0.000203291,0.0001801532,0.06335077,0.09154779,0.003853808,0.003340172,0.835174],"study_design_scores_gemma":[0.00001985818,0.0001036105,0.001983478,0.00001037424,0.0000261141,0.0001646178,0.00005002174,0.9626455,0.02829366,0.004683563,0.001994618,0.00002458814],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02944787,0.0002216629,0.9675098,0.0000595223,0.00002662997,0.00003578912,0.000160597,0.001860084,0.0006780208],"genre_scores_gemma":[0.4482489,0.00040559,0.5417877,0.0001459889,0.0001288703,0.0001637991,0.001723008,0.0007943782,0.006601762],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003312098,"threshold_uncertainty_score":0.006585658,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2126410803","doi":"10.1109/icme.2001.1237829","title":"A music similarity function based on signal analysis","year":2001,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":313,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"PQ Corporation (Canada)","funders":"","keywords":"Similarity (geometry); Computer science; Function (biology); Speech recognition; SIGNAL (programming language); Pattern recognition (psychology); Artificial intelligence; Image (mathematics)","authors":[{"name":"Barry K. Logan","is_ca":true},{"name":"Ariel Salomon","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0274103455880917,"gpt":0.2320162935495742,"spread":0.2046059479614825,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009692167,0.0007921635,0.001105335,0.004894141,0.0004659533,0.001478535,0.001105022,0.001109835,0.003952046],"category_scores_gemma":[0.005668322,0.0001885767,0.0007358435,0.003234149,0.0006830534,0.002151792,0.001243189,0.0007429307,0.003884827],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004005097,"about_ca_system_score_gemma":0.0003680715,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0004911707,"about_ca_topic_score_gemma":0.0004079719,"domain_scores_codex":[0.9975173,0.0003498305,0.0001541109,0.0003094403,0.001585622,0.00008371362],"domain_scores_gemma":[0.9979305,0.0006280692,0.0002539554,0.0002735123,0.0007781623,0.0001357352],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0004228935,0.0002143502,0.002690846,0.0002856133,0.0001379377,0.0001363828,0.00009336713,0.009355049,0.08599064,0.01000934,0.00478395,0.8858796],"study_design_scores_gemma":[0.0001459244,0.001746804,0.01884064,0.00009597449,0.0002181023,0.003147653,0.0001480868,0.8220903,0.0940638,0.03040957,0.02884034,0.0002529421],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01954512,0.0007201465,0.974646,0.0000750709,0.0001769355,0.0001438285,0.0002838802,0.001616985,0.002792091],"genre_scores_gemma":[0.2582121,0.0006435298,0.7346584,0.0001785656,0.0006143415,0.0003589202,0.001448575,0.0003057503,0.003579827],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004894141,"threshold_uncertainty_score":0.01322097,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2295991281","doi":"10.5281/zenodo.1414969","title":"Learning Features From Music Audio With Deep Belief Networks.","year":2010,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Music and Audio Processing","field":"Computer Science","cited_by":259,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal","funders":"","keywords":"Computer science; Feature extraction; Artificial intelligence; Support vector machine; Classifier (UML); Deep belief network; Mel-frequency cepstrum; Task (project management); Speech recognition; Pattern recognition (psychology); Audio signal processing; Frame (networking); Deep learning; Audio signal; Speech coding","authors":[{"name":"Philippe Hamel","is_ca":true},{"name":"Douglas Eck","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01368831642541261,"gpt":0.2031335805673518,"spread":0.1894452641419392,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000436656,0.001216122,0.0009616054,0.0008025503,0.0003435975,0.001098844,0.001516271,0.001301227,0.004750612],"category_scores_gemma":[0.003024026,0.0007782599,0.0008363808,0.001249862,0.0003637331,0.001601121,0.001110429,0.002538357,0.001672955],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005195995,"about_ca_system_score_gemma":0.0005163532,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008386661,"about_ca_topic_score_gemma":0.01378441,"domain_scores_codex":[0.9998197,0.00003398485,0.000009124394,0.00005107396,0.00004218457,0.00004385908],"domain_scores_gemma":[0.999461,0.0002619683,0.00004696236,0.00006855653,0.0001189705,0.00004257364],"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":[0.0007326822,0.0002825281,0.00188839,0.0002399962,0.0002854799,0.0001271146,0.00006740564,0.304852,0.01050064,0.007821428,0.03000649,0.6431959],"study_design_scores_gemma":[0.00001841008,0.00001995809,0.0001927042,0.000009512952,0.00002248,0.0000112507,0.000008635364,0.991874,0.001765446,0.005136236,0.0009353307,0.000005987306],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03413216,0.002372386,0.9500954,0.001184634,0.0006426453,0.00007392225,0.001538151,0.004982349,0.004978231],"genre_scores_gemma":[0.6720709,0.001345162,0.3041699,0.0006029953,0.0005184025,0.00013899,0.005432027,0.0005891907,0.01513252],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.008386661,"threshold_uncertainty_score":0.01667571,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2037202300","doi":"10.1353/lan.2011.0016","title":"The Surfeit of the Stimulus: Analytic Biases Filter Lexical Statistics in Turkish Laryngeal Alternations","year":2011,"lang":"en","type":"article","venue":"Language","topic":"Music and Audio Processing","field":"Computer Science","cited_by":224,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Phonotactics; Lexicon; Turkish; Linguistics; Vowel harmony; Alternation (linguistics); Psychology; Phrase; Phonology; Computer science; Natural language processing","authors":[{"name":"Michael Becker","is_ca":true},{"name":"Nihan Ketrez","is_ca":true},{"name":"Andrew Nevins","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.036742338371428,"gpt":0.2724039918726861,"spread":0.2356616535012581,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002159629,0.0002926973,0.0005280849,0.0006366565,0.000316541,0.001278163,0.0004302147,0.000593964,0.001526255],"category_scores_gemma":[0.03027279,0.0004606869,0.0002641724,0.0004414296,0.001126505,0.001532789,0.00141288,0.0007254808,0.000301403],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004520596,"about_ca_system_score_gemma":0.0003348319,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00130588,"about_ca_topic_score_gemma":0.0009301495,"domain_scores_codex":[0.998212,0.0005850295,0.0001428362,0.0005217624,0.0003882113,0.0001502966],"domain_scores_gemma":[0.987619,0.008378696,0.001469237,0.001677451,0.0005232315,0.0003323998],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.002046767,0.0002294559,0.1017553,0.0003609838,0.0001316646,0.0005417758,0.006450647,0.005860819,0.7327883,0.007875057,0.0008485018,0.1411107],"study_design_scores_gemma":[0.0001810933,0.0009507169,0.7241201,0.00007826401,0.0002370784,0.001817025,0.002200585,0.09545103,0.1311207,0.03991615,0.003671793,0.0002555303],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9908401,0.00004426436,0.00652665,0.0000645031,0.000009429712,0.00002032846,0.00005331607,0.0001340311,0.002307353],"genre_scores_gemma":[0.9972821,0.00002132478,0.002266156,0.00004328231,0.00000439154,0.0000179211,0.00007970521,0.00008252955,0.0002025893],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.002159629,"threshold_uncertainty_score":0.01142132,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2162124943","doi":"","title":"Automatic Generation of Social Tags for Music Recommendation","year":2007,"lang":"en","type":"article","venue":"neural information processing systems","topic":"Music and Audio Processing","field":"Computer Science","cited_by":184,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal; CMC Microsystems (Canada)","funders":"","keywords":"Recommender system; Computer science; Information retrieval; Set (abstract data type); Baseline (sea); World Wide Web; Resource (disambiguation); Space (punctuation); Social web; Component (thermodynamics); Social media","authors":[{"name":"Douglas Eck","is_ca":true},{"name":"Paul Lamere","is_ca":false},{"name":"Thierry Bertin-Mahieux","is_ca":false},{"name":"Stephen Green","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06320323338863898,"gpt":0.294099101272341,"spread":0.230895867883702,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008098643,0.0007873744,0.000594289,0.00245588,0.0007019604,0.0009317625,0.0009474471,0.00109874,0.002134195],"category_scores_gemma":[0.004356659,0.0003603437,0.0007173113,0.001774466,0.000326475,0.001243797,0.0007483168,0.0008692308,0.003539399],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007322888,"about_ca_system_score_gemma":0.0007542986,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007290555,"about_ca_topic_score_gemma":0.02383797,"domain_scores_codex":[0.9989443,0.0002696712,0.0000537621,0.0002891798,0.0003174701,0.0001256739],"domain_scores_gemma":[0.9975771,0.0008728913,0.0001768653,0.0004407816,0.0008279922,0.0001043925],"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":[0.0005868218,0.0005039651,0.02248252,0.0002410658,0.0002204461,0.0002768603,0.0003943761,0.02534584,0.0415266,0.004889013,0.02922503,0.8743074],"study_design_scores_gemma":[0.00006069087,0.0001472151,0.009521491,0.00003429352,0.00009708119,0.000248689,0.0002157333,0.9234076,0.04138378,0.0106063,0.01421578,0.00006142438],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1295759,0.0008535698,0.846418,0.0005157756,0.0004194652,0.000398199,0.00359002,0.009299411,0.008929669],"genre_scores_gemma":[0.5972756,0.0003324725,0.384872,0.0001866982,0.0002384157,0.0002853884,0.007128368,0.0003756392,0.009305431],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007290555,"threshold_uncertainty_score":0.01449621,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2106339924","doi":"10.1109/icassp.2000.859336","title":"Speech/music discrimination for multimedia applications","year":2002,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":181,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Multimedia; Speech recognition","authors":[{"name":"Khaled El-Maleh","is_ca":true},{"name":"Mark D. Klein","is_ca":true},{"name":"G. Petrucci","is_ca":true},{"name":"P. Kabal","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05160292578650467,"gpt":0.2628288248486828,"spread":0.2112258990621781,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006851092,0.0006495385,0.0005137128,0.001698092,0.0004082707,0.00075636,0.0005650295,0.0008880994,0.007658463],"category_scores_gemma":[0.002298771,0.0001653875,0.0002964004,0.0009104576,0.0001787798,0.0008498423,0.0004214437,0.0004528167,0.002639305],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003415274,"about_ca_system_score_gemma":0.0002857143,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008225693,"about_ca_topic_score_gemma":0.001228737,"domain_scores_codex":[0.9995178,0.00009679677,0.00002808433,0.0000884401,0.0002138886,0.00005500417],"domain_scores_gemma":[0.9988594,0.0004795384,0.00006176323,0.00008346473,0.0004049489,0.0001107833],"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":[0.00155158,0.0002104379,0.003650723,0.0002440953,0.00003152016,0.0001795674,0.00003204591,0.003321015,0.3884096,0.0008435679,0.003567481,0.5979583],"study_design_scores_gemma":[0.0002047909,0.001732262,0.04076867,0.00009039896,0.0002008787,0.001679228,0.0001575337,0.3010417,0.6301562,0.003658445,0.02020161,0.0001083029],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"methods","genre_scores_codex":[0.5444233,0.008923497,0.4188551,0.001183782,0.0005560722,0.0005237555,0.001210185,0.008025791,0.01629843],"genre_scores_gemma":[0.7817074,0.001346037,0.207473,0.0003120016,0.0002623928,0.0001351307,0.00146598,0.0002413856,0.007056609],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.007658463,"threshold_uncertainty_score":0.0256201,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1801907668","doi":"10.5281/zenodo.1416157","title":"Automatic Genre Classification Using Large High-Level Musical Feature Sets.","year":2004,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":153,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Computer science; MIDI; Artificial intelligence; Hierarchy; Pattern recognition (psychology); Feature (linguistics); Classifier (UML); Music information retrieval; Feature extraction; Embedding; Rhythm; Speech recognition; Musical; Linguistics","authors":[{"name":"Cory McKay","is_ca":true},{"name":"Ichiro Fujinaga","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0597030218101955,"gpt":0.2884801686068033,"spread":0.2287771467966078,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008439842,0.001090967,0.001452068,0.004343466,0.0008124494,0.001406088,0.001397976,0.001267805,0.005988601],"category_scores_gemma":[0.004003907,0.0004729737,0.001337495,0.002814815,0.0003196458,0.001781776,0.001498975,0.001499909,0.005174298],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002414102,"about_ca_system_score_gemma":0.0004124665,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002427081,"about_ca_topic_score_gemma":0.005891139,"domain_scores_codex":[0.9992716,0.0001348905,0.00005537194,0.0001677165,0.0002536422,0.0001168571],"domain_scores_gemma":[0.9987306,0.0004217776,0.00005986713,0.000310715,0.0003733565,0.0001036831],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0007308753,0.0003675447,0.003692027,0.0002543888,0.0002253227,0.0002255113,0.00009150456,0.004540795,0.08606872,0.001606591,0.04284082,0.8593559],"study_design_scores_gemma":[0.0003398171,0.0005075808,0.03852287,0.0001358385,0.0005165694,0.001016641,0.0005505627,0.8195062,0.07324119,0.02426975,0.04122094,0.0001720138],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.08664647,0.003322853,0.8731998,0.0009088902,0.001495024,0.0003870433,0.008364837,0.01405215,0.01162294],"genre_scores_gemma":[0.3467439,0.001340006,0.6091505,0.0004577898,0.0008577235,0.0003507994,0.02721814,0.0007778153,0.01310323],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005988601,"threshold_uncertainty_score":0.02003384,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2395935897","doi":"10.5281/zenodo.1418318","title":"Audio Chord Recognition With Recurrent Neural Networks.","year":2013,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Music and Audio Processing","field":"Computer Science","cited_by":145,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal","funders":"","keywords":"Computer science; Chord (peer-to-peer); Artificial neural network; Recurrent neural network; Speech recognition; Artificial intelligence; Time delay neural network; Pattern recognition (psychology); Machine learning","authors":[{"name":"Nicolas Boulanger-Lewandowski","is_ca":true},{"name":"Yoshua Bengio","is_ca":true},{"name":"Pascal Vincent","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.03494166430596236,"gpt":0.2214052918098112,"spread":0.1864636275038488,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000363195,0.0007737055,0.0005500889,0.0006462421,0.0002458418,0.000956925,0.001302512,0.0007264692,0.007999709],"category_scores_gemma":[0.002021017,0.0003452517,0.0005358921,0.0007052033,0.0002216274,0.00115396,0.0008972671,0.001098252,0.004254657],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002975959,"about_ca_system_score_gemma":0.0002867116,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005431547,"about_ca_topic_score_gemma":0.009125577,"domain_scores_codex":[0.9997988,0.00002918328,0.00001422057,0.00006245813,0.0000580659,0.00003728845],"domain_scores_gemma":[0.9996119,0.00009944772,0.00003025767,0.00008481264,0.0001428265,0.00003075146],"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":[0.0006210158,0.0001263804,0.001004954,0.0003509542,0.0002202349,0.0002387077,0.00006828103,0.02999955,0.06048355,0.006127634,0.04857865,0.85218],"study_design_scores_gemma":[0.00004354589,0.0001329171,0.001165506,0.00003938405,0.0001181389,0.0001393191,0.00004963228,0.9339439,0.03717152,0.01083212,0.01632736,0.00003664554],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02607447,0.004488197,0.9381759,0.0007746847,0.001976482,0.0001150583,0.002336565,0.01194527,0.01411337],"genre_scores_gemma":[0.5431477,0.002473409,0.4033999,0.0004299933,0.000953195,0.0001169426,0.01039005,0.0008937069,0.03819513],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007999709,"threshold_uncertainty_score":0.02676171,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1981219830","doi":"10.1145/2601097.2601110","title":"Exploratory font selection using crowdsourced attributes","year":2014,"lang":"en","type":"article","venue":"ACM Transactions on Graphics","topic":"Music and Audio Processing","field":"Computer Science","cited_by":144,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"Natural Sciences and Engineering Research Council of Canada; Canadian Institute for Advanced Research; Adobe Systems","keywords":"Font; Computer science; Selection (genetic algorithm); Similarity (geometry); Crowdsourcing; Information retrieval; Artificial intelligence; Metric (unit); Human–computer interaction; Natural language processing; World Wide Web","authors":[{"name":"Peter O’Donovan","is_ca":true},{"name":"Jānis Lībeks","is_ca":true},{"name":"Aseem Agarwala","is_ca":false},{"name":"Aaron Hertzmann","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04275869709030147,"gpt":0.2576552930373095,"spread":0.214896595947008,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003232789,0.002330616,0.001644264,0.003445603,0.001131783,0.00224653,0.002269529,0.001575376,0.007520917],"category_scores_gemma":[0.02277584,0.0005044388,0.001448103,0.001949587,0.0007521462,0.003584004,0.004485912,0.0012045,0.00341315],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008180782,"about_ca_system_score_gemma":0.000812071,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003088526,"about_ca_topic_score_gemma":0.005654335,"domain_scores_codex":[0.9952639,0.001859038,0.0002026032,0.0009280666,0.00159174,0.0001545802],"domain_scores_gemma":[0.9857782,0.009470516,0.0005370659,0.00224547,0.001415312,0.0005534105],"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":[0.004082555,0.001924568,0.02082598,0.002964831,0.0004203622,0.00117892,0.006913675,0.04312614,0.04380932,0.00730264,0.05168235,0.8157687],"study_design_scores_gemma":[0.0007322929,0.001596738,0.02379705,0.0006313807,0.0002984178,0.001223446,0.006724475,0.7034979,0.04485266,0.06327277,0.1526803,0.0006925257],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2762094,0.001998606,0.6369928,0.0008641979,0.0004923432,0.002618704,0.009619514,0.03687477,0.03432975],"genre_scores_gemma":[0.5145292,0.0003412876,0.4648024,0.0004623547,0.0001256315,0.001697483,0.008047695,0.001527597,0.008466337],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007520917,"threshold_uncertainty_score":0.02515996,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3109961563","doi":"10.3390/math8122133","title":"CLSTM: Deep Feature-Based Speech Emotion Recognition Using the Hierarchical ConvLSTM Network","year":2020,"lang":"en","type":"article","venue":"Mathematics","topic":"Music and Audio Processing","field":"Computer Science","cited_by":142,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Computer science; Artificial intelligence; Deep learning; Feature (linguistics); Block (permutation group theory); Machine learning; State (computer science); Pattern recognition (psychology)","authors":[{"name":"Mustaqeem Mustaqeem","is_ca":false},{"name":"Soonil Kwon","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06164806315143793,"gpt":0.2596169155525182,"spread":0.1979688524010803,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002198634,0.000717083,0.0003179283,0.0002938501,0.0001586208,0.0003390919,0.0009258576,0.0005528426,0.00312358],"category_scores_gemma":[0.00045807,0.0002144732,0.0004755632,0.0002993511,0.0002321541,0.0007803233,0.0005908191,0.0007832033,0.001204848],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004365651,"about_ca_system_score_gemma":0.0004891757,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006570281,"about_ca_topic_score_gemma":0.009016883,"domain_scores_codex":[0.9998854,0.00001074759,0.000006736766,0.00004196892,0.00003267901,0.00002235889],"domain_scores_gemma":[0.9999235,0.00001626907,0.00001112626,0.00001020254,0.00003026871,0.000008667395],"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":[0.0003784232,0.0002551995,0.001562013,0.0003453649,0.0001701013,0.0003612496,0.0001407728,0.119391,0.15315,0.004548878,0.02182152,0.6978756],"study_design_scores_gemma":[0.0000138393,0.00008762832,0.0007521791,0.00001298121,0.00002245919,0.00005745813,0.00001243813,0.9798626,0.01488009,0.001726666,0.002557964,0.00001366932],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.05792226,0.001644936,0.9208341,0.0004284394,0.0004016598,0.0001280929,0.001010523,0.01157307,0.006056906],"genre_scores_gemma":[0.7996004,0.0007822945,0.1867494,0.0005275633,0.0001180059,0.000190503,0.002499559,0.0003119582,0.009220236],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006570281,"threshold_uncertainty_score":0.01306409,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2090879738","doi":"10.1007/s00530-006-0032-2","title":"Support vector machine active learning for music retrieval","year":2006,"lang":"en","type":"article","venue":"Multimedia Systems","topic":"Music and Audio Processing","field":"Computer Science","cited_by":133,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"York University","funders":"National Science Foundation","keywords":"Computer science; Similarity (geometry); Music information retrieval; Artificial intelligence; Digital audio; Support vector machine; Scheme (mathematics); Active learning (machine learning); Machine learning; Style (visual arts); Information retrieval; Speech recognition; Pattern recognition (psychology); Audio signal; Musical; Mathematics; Speech coding","authors":[{"name":"Michael Mandel","is_ca":true},{"name":"Graham E. Poliner","is_ca":true},{"name":"Daniel P. W. Ellis","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01953620587089179,"gpt":0.2414020996941554,"spread":0.2218658938232636,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001896101,0.0006161536,0.00155767,0.0009996311,0.0004624027,0.001193157,0.001878898,0.001294681,0.003019212],"category_scores_gemma":[0.00626516,0.0004217363,0.0005456114,0.001699842,0.0005240623,0.002612866,0.0007967686,0.001767107,0.001486354],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003881815,"about_ca_system_score_gemma":0.000676572,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002942989,"about_ca_topic_score_gemma":0.002557105,"domain_scores_codex":[0.998862,0.0003866509,0.0001071508,0.0001659182,0.0003973457,0.00008093435],"domain_scores_gemma":[0.9966433,0.001914439,0.0001742949,0.000328687,0.0008829065,0.00005637046],"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":[0.0004509886,0.0002927927,0.0004213103,0.0002246277,0.00008697448,0.00005580313,0.00006885322,0.08212306,0.01158992,0.008149349,0.006946694,0.8895895],"study_design_scores_gemma":[0.00002339862,0.00004958281,0.0001345111,0.000005172401,0.00001714376,0.00002324742,0.00001177209,0.9902847,0.003331723,0.004873257,0.00123627,0.000009151951],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01527022,0.003158442,0.9784759,0.0003587173,0.0001721328,0.00004454542,0.0001246796,0.001298119,0.00109716],"genre_scores_gemma":[0.4926655,0.002138555,0.4865512,0.0002973322,0.0007952445,0.0001961978,0.001009762,0.0002705548,0.01607558],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003019212,"threshold_uncertainty_score":0.01010031,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2119662931","doi":"10.1109/tasl.2011.2118753","title":"Time–Frequency Matrix Feature Extraction and Classification of Environmental Audio Signals","year":2011,"lang":"en","type":"article","venue":"IEEE Transactions on Audio Speech and Language Processing","topic":"Music and Audio Processing","field":"Computer Science","cited_by":125,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Mel-frequency cepstrum; Computer science; Feature extraction; Audio signal; Speech recognition; Pattern recognition (psychology); Artificial intelligence; Audio signal processing; Support vector machine; Feature (linguistics); Time–frequency analysis; Speech coding; Computer vision","authors":[{"name":"Behnaz Ghoraani","is_ca":true},{"name":"Sridhar Krishnan","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01521591215504704,"gpt":0.247987920621622,"spread":0.232772008466575,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002724908,0.0004189179,0.0002723971,0.001100777,0.0001748835,0.0003868172,0.0002426924,0.0003702177,0.001092157],"category_scores_gemma":[0.001359537,0.00009569246,0.0003631825,0.0007194816,0.0001793276,0.000527215,0.0002254808,0.0002995926,0.0005476677],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001548122,"about_ca_system_score_gemma":0.0002226589,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001827543,"about_ca_topic_score_gemma":0.002519874,"domain_scores_codex":[0.999747,0.00003160243,0.00001577382,0.00004781706,0.0001306759,0.00002728546],"domain_scores_gemma":[0.9996389,0.0001172403,0.00004658391,0.00003249864,0.0001508224,0.00001385039],"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":[0.0001689498,0.00008680898,0.003746475,0.0001002795,0.00003181462,0.0002025353,0.0000646605,0.02700927,0.1704989,0.001721463,0.001363072,0.7950058],"study_design_scores_gemma":[0.00002007845,0.0002179505,0.03255482,0.00001967035,0.00004406898,0.0004266274,0.0001523065,0.8761273,0.08252481,0.001692113,0.006181167,0.00003906977],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1171949,0.0003646056,0.8797684,0.0001155338,0.00008112058,0.00007964257,0.0002636662,0.0006046342,0.001527534],"genre_scores_gemma":[0.6107772,0.0005305524,0.3843452,0.0000524325,0.00007405946,0.0001233579,0.0008313034,0.00005496037,0.003210902],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.001827543,"threshold_uncertainty_score":0.003653646,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W56995320","doi":"10.5281/zenodo.1417417","title":"Musical Genre Classification: Is It Worth Pursuing And How Can It Be Improved?","year":2006,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Music and Audio Processing","field":"Computer Science","cited_by":121,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Subjectivity; Similarity (geometry); Musical; Genre analysis; Discipline; Data science; Music information retrieval; Musicology; Linguistics; Artificial intelligence; Natural language processing; Epistemology; Sociology; Social science; Pedagogy; Literature","authors":[{"name":"Cory McKay","is_ca":true},{"name":"Ichiro Fujinaga","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05322688878533093,"gpt":0.2470623955776686,"spread":0.1938355067923377,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004961367,0.001318224,0.002967946,0.004204228,0.001852514,0.006476068,0.003490251,0.002188093,0.02012341],"category_scores_gemma":[0.02417556,0.0004172398,0.001353609,0.004051739,0.001191812,0.008778988,0.002008749,0.00253384,0.02664444],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007855211,"about_ca_system_score_gemma":0.001487391,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007509409,"about_ca_topic_score_gemma":0.01034625,"domain_scores_codex":[0.9976693,0.0006384713,0.0002409286,0.0005285477,0.0007059703,0.0002168001],"domain_scores_gemma":[0.9859188,0.002324894,0.0005426205,0.003441695,0.006524885,0.001247207],"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":[0.0004551665,0.0002090561,0.008827232,0.0004933377,0.0001108788,0.00005409882,0.0001589516,0.001231913,0.009147104,0.006039081,0.08696456,0.8863087],"study_design_scores_gemma":[0.000713423,0.00111912,0.08596422,0.002071073,0.0009972586,0.002344247,0.006016155,0.2952836,0.03039301,0.3025377,0.2718172,0.000742965],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07682005,0.0204147,0.7332065,0.07929359,0.01282696,0.0008330936,0.006041896,0.02249964,0.04806363],"genre_scores_gemma":[0.2336954,0.01069153,0.6938047,0.006810068,0.007517221,0.0003720809,0.01306244,0.00243414,0.03161247],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.02012341,"threshold_uncertainty_score":0.06731957,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W4213366994","doi":"10.1109/lsp.2022.3150258","title":"CNN-RNN and Data Augmentation Using Deep Convolutional Generative Adversarial Network for Environmental Sound Classification","year":2022,"lang":"en","type":"article","venue":"IEEE Signal Processing Letters","topic":"Music and Audio Processing","field":"Computer Science","cited_by":118,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Simon Fraser University","funders":"Kids Brain Health Network","keywords":"Computer science; Artificial intelligence; Deep learning; Convolutional neural network; Recurrent neural network; Normalization (sociology); Transfer of learning; Feature extraction; Pattern recognition (psychology); Generative adversarial network; Feature (linguistics); Feature learning; Machine learning; Spectrogram; Generative grammar; Artificial neural network","authors":[{"name":"Behnaz Bahmei","is_ca":true},{"name":"Elina Birmingham","is_ca":true},{"name":"Siamak Arzanpour","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.07475645323796545,"gpt":0.2863002993979042,"spread":0.2115438461599388,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006460958,0.0009808864,0.0006058802,0.0005346725,0.0002145324,0.0003877124,0.0009269548,0.0006347485,0.001618777],"category_scores_gemma":[0.00119521,0.0002902051,0.0008099669,0.0006656364,0.0004564954,0.0007578639,0.0009193794,0.001190559,0.0005814052],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005239834,"about_ca_system_score_gemma":0.0005533245,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006514094,"about_ca_topic_score_gemma":0.007120866,"domain_scores_codex":[0.9996058,0.00007418611,0.00002142996,0.0001292125,0.0001094203,0.00005995126],"domain_scores_gemma":[0.9996866,0.0001097765,0.00003126434,0.00007899683,0.0000781679,0.00001526547],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002546455,0.0001675171,0.002439467,0.00009811731,0.0001073784,0.0002288099,0.00005348647,0.6437798,0.02175371,0.004168205,0.005421068,0.3215278],"study_design_scores_gemma":[0.000002572133,0.00001934948,0.0002323179,0.000003207996,0.000006565653,0.00001691946,0.000003908062,0.995041,0.003533866,0.0006343884,0.0005005901,0.000005380874],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07777866,0.0009756419,0.9110161,0.0003856188,0.0002311868,0.00009807911,0.000502526,0.004516687,0.004495495],"genre_scores_gemma":[0.8033028,0.0004307681,0.1865275,0.0002948698,0.00007250442,0.0001721958,0.002257281,0.0001638308,0.006778199],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006514094,"threshold_uncertainty_score":0.01295239,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2296557957","doi":"10.5281/zenodo.1417546","title":"An Expert Ground Truth Set For Audio Chord Recognition And Music Analysis.","year":2011,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Music and Audio Processing","field":"Computer Science","cited_by":114,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"","keywords":"Chord (peer-to-peer); Computer science; Speech recognition; Music information retrieval; Phrase; USable; Ground truth; Melody; Natural language processing; Scope (computer science); Chart; Artificial intelligence; Information retrieval; Musical; Multimedia; Database; Mathematics","authors":[{"name":"John Burgoyne","is_ca":true},{"name":"Jonathan Wild","is_ca":true},{"name":"Ichiro Fujinaga","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1466874330866852,"gpt":0.2704199464206599,"spread":0.1237325133339746,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003606564,0.002305602,0.001662319,0.0059169,0.001667139,0.002206105,0.003371429,0.003789027,0.01973913],"category_scores_gemma":[0.0140276,0.0007979502,0.001556161,0.003207489,0.001094624,0.002519862,0.002485933,0.001704118,0.01384732],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009648913,"about_ca_system_score_gemma":0.002378308,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01607568,"about_ca_topic_score_gemma":0.02988841,"domain_scores_codex":[0.9956706,0.0008533579,0.0003197773,0.001506662,0.001310454,0.0003392038],"domain_scores_gemma":[0.991593,0.002356244,0.0002345179,0.002549774,0.002918638,0.0003478106],"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":[0.001649925,0.0009326501,0.005239265,0.001934247,0.0005697564,0.0009608539,0.0002171747,0.01965763,0.03318153,0.004061741,0.2527381,0.678857],"study_design_scores_gemma":[0.001136418,0.00142164,0.04122708,0.001322579,0.001209942,0.004083011,0.001635772,0.4909066,0.1490322,0.03202304,0.2756397,0.0003619785],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"dataset","genre_scores_codex":[0.09227863,0.0079174,0.6013478,0.001349449,0.002108185,0.001537128,0.2305552,0.03652632,0.02637992],"genre_scores_gemma":[0.2121205,0.00102737,0.2920134,0.0003968275,0.0002480549,0.0006327535,0.4812095,0.00115862,0.01119304],"genre_candidate":"dataset","genre_consensus":null,"teacher_disagreement_score":0.01973913,"threshold_uncertainty_score":0.06603396,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2064082346","doi":"10.1109/icassp.2010.5495651","title":"Phone recognition using Restricted Boltzmann Machines","year":2010,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":108,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"TIMIT; Hidden Markov model; Computer science; Phone; Conditional independence; Boltzmann machine; Independence (probability theory); Set (abstract data type); Speech recognition; Generative model; Artificial intelligence; Restricted Boltzmann machine; Markov model; Generative grammar; Machine learning; Pattern recognition (psychology); Markov chain; Artificial neural network; Mathematics","authors":[{"name":"Abdelrahman Mohamed","is_ca":true},{"name":"Geoffrey E. Hinton","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0326160043459392,"gpt":0.2635966483133135,"spread":0.2309806439673743,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008751366,0.0006675766,0.001110033,0.0007146866,0.0003275065,0.0009688524,0.001149575,0.001142671,0.00458791],"category_scores_gemma":[0.003798445,0.0006102833,0.001074671,0.0007381598,0.0003883427,0.001632647,0.001291139,0.001357666,0.004566985],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003455871,"about_ca_system_score_gemma":0.00048626,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002185129,"about_ca_topic_score_gemma":0.00233105,"domain_scores_codex":[0.9990682,0.0003376459,0.00005266757,0.0002439071,0.000215114,0.00008256327],"domain_scores_gemma":[0.9991763,0.0004330713,0.00005324834,0.0001802533,0.0001235626,0.00003351912],"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":[0.0002118432,0.00008774365,0.001571872,0.0001971457,0.0002163136,0.0002003569,0.0001468822,0.3583078,0.02012385,0.01808208,0.007485845,0.5933683],"study_design_scores_gemma":[0.00000927711,0.00001933818,0.0002749778,0.00001246806,0.00001116117,0.00007213307,0.00001441195,0.9765266,0.004124033,0.016626,0.00228912,0.00002042059],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01189723,0.0007135334,0.9805367,0.0001850227,0.0001371007,0.00003699368,0.0002549343,0.003953072,0.002285441],"genre_scores_gemma":[0.4846609,0.0009822759,0.5028925,0.0003732711,0.0001475825,0.0002427144,0.001768751,0.0005269659,0.008405129],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00458791,"threshold_uncertainty_score":0.01534814,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W179394430","doi":"","title":"Disambiguating Music Emotion Using Software Agents.","year":2004,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":105,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa","funders":"","keywords":"Lyrics; Computer science; Heuristics; Active listening; Ambiguity; Metadata; Focus (optics); Software; Annotation; Artificial intelligence; Natural language processing; Cognitive psychology; Human–computer interaction; Psychology; World Wide Web; Communication","authors":[{"name":"Yang Dan","is_ca":true},{"name":"Won‐Sook Lee","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0664556378404012,"gpt":0.2801708730104419,"spread":0.2137152351700407,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002527395,0.0008888388,0.0005402587,0.0007321144,0.0008068521,0.00217511,0.001036122,0.001190772,0.003133175],"category_scores_gemma":[0.01094891,0.0004140826,0.0006498351,0.0004299677,0.0006783889,0.002035588,0.001775375,0.00110822,0.00124921],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004641912,"about_ca_system_score_gemma":0.0005924778,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001828044,"about_ca_topic_score_gemma":0.002872857,"domain_scores_codex":[0.998763,0.0005101491,0.0001031084,0.0003027284,0.0002500199,0.00007102607],"domain_scores_gemma":[0.9961069,0.002322203,0.0004067623,0.0004532616,0.0004800291,0.0002307206],"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":[0.002364255,0.001312143,0.03062027,0.0008194824,0.0004184114,0.0007600355,0.008014432,0.03214667,0.1361149,0.03387312,0.01330901,0.7402472],"study_design_scores_gemma":[0.0003884923,0.000579762,0.01198467,0.0001196652,0.0003346677,0.0004332665,0.003139114,0.8075284,0.05291555,0.05935734,0.06306763,0.0001514101],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1770355,0.0004265443,0.7960274,0.001266644,0.0002794597,0.0008723473,0.0003217532,0.008594705,0.01517561],"genre_scores_gemma":[0.4013658,0.0001758694,0.5873128,0.0003025603,0.00005842424,0.0006939046,0.0005897582,0.0002649707,0.009235982],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003133175,"threshold_uncertainty_score":0.01336634,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W123034841","doi":"10.1016/b978-0-12-381460-9.00002-x","title":"Musical Timbre Perception","year":2012,"lang":"en","type":"book-chapter","venue":"Elsevier eBooks","topic":"Music and Audio Processing","field":"Computer Science","cited_by":101,"is_retracted":false,"has_abstract":false,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"Core Research for Evolutional Science and Technology","keywords":"Timbre; Musical; Perception; Art; Psychology; Communication; Speech recognition; Visual arts; Computer science; Neuroscience","authors":[{"name":"Stephen McAdams","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02194690475241409,"gpt":0.2378707900940175,"spread":0.2159238853416034,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001907997,0.0006710197,0.0004035441,0.0007256438,0.0003566087,0.002279364,0.0006688748,0.0008698774,0.1085077],"category_scores_gemma":[0.0003793254,0.0002897592,0.0002705166,0.0006841834,0.0004366057,0.001559808,0.00104414,0.0006689529,0.04607846],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002596484,"about_ca_system_score_gemma":0.0002528949,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000837819,"about_ca_topic_score_gemma":0.001241787,"domain_scores_codex":[0.9998989,0.000005516089,0.000004035695,0.00002646867,0.00005835083,0.000006593368],"domain_scores_gemma":[0.9999307,0.00001643962,0.000002883877,0.00001299491,0.00002338378,0.00001361205],"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":[0.00003741948,0.00001952726,0.0001093411,0.0001596418,0.00000686497,0.00005791681,0.0001300871,0.0002272868,0.03350566,0.008631885,0.03434715,0.9227673],"study_design_scores_gemma":[0.00001795887,0.0001152728,0.005934414,0.0003540314,0.00002681737,0.001464636,0.0003002631,0.003897081,0.01762656,0.02887449,0.9413467,0.00004182358],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"other","genre_gemma":"other","genre_scores_codex":[0.006417373,0.02399374,0.1672724,0.0008693538,0.00166907,0.00005287591,0.0005156636,0.003445407,0.7957641],"genre_scores_gemma":[0.04187556,0.01391021,0.03059957,0.0004356304,0.0004996638,0.00004956229,0.0006580863,0.0006258429,0.911346],"genre_candidate":"other","genre_consensus":"other","teacher_disagreement_score":0.1085077,"threshold_uncertainty_score":0.3629943,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2109573991","doi":"10.1155/2010/451695","title":"Audio Signal Processing Using Time-Frequency Approaches: Coding, Classification, Fingerprinting, and Watermarking","year":2010,"lang":"en","type":"article","venue":"EURASIP Journal on Advances in Signal Processing","topic":"Music and Audio Processing","field":"Computer Science","cited_by":97,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Computer science; Digital watermarking; Audio signal; Digital audio; Speech coding; Audio signal processing; Speech recognition; Coding (social sciences); Psychoacoustics; Signal processing; Artificial intelligence; Digital signal processing; Perception; Computer hardware","authors":[{"name":"Karthikeyan Umapathy","is_ca":true},{"name":"Behnaz Ghoraani","is_ca":true},{"name":"Sridhar Krishnan","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.04103532990992343,"gpt":0.2835590867445217,"spread":0.2425237568345983,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005889306,0.0007059641,0.0004531339,0.001547629,0.0002991214,0.001062275,0.0005723256,0.00156044,0.001543762],"category_scores_gemma":[0.001453021,0.0002045075,0.0004341897,0.002547383,0.001043658,0.001595451,0.0005134485,0.0007548037,0.001014015],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002997269,"about_ca_system_score_gemma":0.0002780612,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001029313,"about_ca_topic_score_gemma":0.0007787009,"domain_scores_codex":[0.9995381,0.00007967171,0.00002629204,0.00008185586,0.0002484185,0.00002562515],"domain_scores_gemma":[0.9995209,0.0001824257,0.00007170136,0.00006415712,0.0001429656,0.00001786046],"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":[0.0001403348,0.00007255629,0.0005420704,0.0005676706,0.00004187679,0.0001779528,0.0001496985,0.01410817,0.0908326,0.03297355,0.003010368,0.8573832],"study_design_scores_gemma":[0.00006112108,0.001044155,0.005554676,0.000525084,0.0002232078,0.004786387,0.0003696793,0.5747241,0.1846647,0.1153997,0.1123701,0.000277033],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.009767645,0.02081635,0.961947,0.0005199982,0.0003858852,0.00006969971,0.00005148203,0.0003809039,0.006061001],"genre_scores_gemma":[0.2271005,0.04417158,0.7088805,0.0006119869,0.002032675,0.0001719577,0.0002725953,0.000106594,0.01665151],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.00156044,"threshold_uncertainty_score":0.005164385,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2162911105","doi":"10.1109/icassp.2013.6638244","title":"High-dimensional sequence transduction","year":2013,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":96,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal","funders":"","keywords":"Polyphony; Sequence (biology); Computer science; Probabilistic logic; Speech recognition; Notation; Noise (video); Mode (computer interface); Artificial neural network; Recurrent neural network; Transduction (biophysics); Algorithm; Artificial intelligence; Mathematics; Arithmetic","authors":[{"name":"Nicolas Boulanger-Lewandowski","is_ca":true},{"name":"Yoshua Bengio","is_ca":true},{"name":"Pascal Vincent","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02160387256074852,"gpt":0.2278007704775471,"spread":0.2061968979167986,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001346921,0.0008435264,0.0007884341,0.0004789197,0.0003672492,0.0008662344,0.001216245,0.001364332,0.004205697],"category_scores_gemma":[0.00459237,0.0003464435,0.0008798837,0.000771147,0.00107565,0.002048721,0.001818468,0.001376277,0.001982006],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000436588,"about_ca_system_score_gemma":0.0005001823,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005711943,"about_ca_topic_score_gemma":0.0005397579,"domain_scores_codex":[0.9988055,0.0003721109,0.0000691591,0.0004088875,0.0002690013,0.00007527717],"domain_scores_gemma":[0.9982925,0.0009835358,0.0001321116,0.000292697,0.0002376612,0.0000615368],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0002939417,0.000250947,0.001149303,0.0003722276,0.000110594,0.0004829377,0.0003494805,0.4986211,0.08550141,0.05055045,0.003241515,0.3590761],"study_design_scores_gemma":[0.00001327949,0.0001256435,0.0002421337,0.00001122065,0.00001105905,0.0001203993,0.00003750889,0.9604539,0.01177707,0.0258211,0.001369176,0.00001750701],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02702953,0.0001612502,0.9697232,0.0001693596,0.00005141464,0.00004141278,0.0001376056,0.00133609,0.001350183],"genre_scores_gemma":[0.6789021,0.0003525243,0.312146,0.0003526713,0.000108174,0.0002259894,0.001377512,0.0003331391,0.006201907],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004205697,"threshold_uncertainty_score":0.0140695,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2108672713","doi":"10.1145/1631272.1631393","title":"Improving automatic music tag annotation using stacked generalization of probabilistic SVM outputs","year":2009,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":96,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Victoria","funders":"","keywords":"Annotation; Computer science; Probabilistic logic; Generalization; Support vector machine; Music information retrieval; Information retrieval; Recommender system; Artificial intelligence; Scheme (mathematics); Natural language processing; Machine learning; Speech recognition; Musical","authors":[{"name":"Steven R. Ness","is_ca":true},{"name":"Anthony Theocharis","is_ca":true},{"name":"George Tzanetakis","is_ca":true},{"name":"Luís Gustavo Martins","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02967875432838947,"gpt":0.2574788825945556,"spread":0.2278001282661661,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00231303,0.002128183,0.001851147,0.001344027,0.0007240112,0.001193178,0.00191111,0.001985661,0.002172995],"category_scores_gemma":[0.006506134,0.0006873554,0.001396879,0.001483033,0.0004251324,0.002681584,0.001326802,0.002365291,0.003045026],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006576643,"about_ca_system_score_gemma":0.001156523,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0108856,"about_ca_topic_score_gemma":0.01480583,"domain_scores_codex":[0.998657,0.0003530575,0.00009234269,0.0004428832,0.0002818328,0.0001728416],"domain_scores_gemma":[0.9961621,0.001689329,0.0002498518,0.0008137158,0.000952424,0.0001326195],"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":[0.0006888712,0.0004957787,0.006333814,0.0001087473,0.0002672416,0.0001241418,0.00014088,0.1206176,0.01953273,0.0008374748,0.01188702,0.8389658],"study_design_scores_gemma":[0.0000218553,0.00006390999,0.001441913,0.000007667812,0.00004930887,0.00003651316,0.0000332303,0.9913927,0.004944746,0.001396605,0.0005949736,0.00001652337],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.1793728,0.001312821,0.7897946,0.0004551715,0.0002374478,0.00008557364,0.0009956653,0.02505125,0.002694659],"genre_scores_gemma":[0.7373196,0.0003195847,0.2506263,0.000408349,0.00020536,0.0001110536,0.005663702,0.0005996889,0.004746398],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.0108856,"threshold_uncertainty_score":0.02164447,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2406196141","doi":"10.5281/zenodo.1418237","title":"Temporal Pooling And Multiscale Learning For Automatic Annotation And Ranking Of Music Audio.","year":2011,"lang":"en","type":"article","venue":"Zenodo (CERN European Organization for Nuclear Research)","topic":"Music and Audio Processing","field":"Computer Science","cited_by":95,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal","funders":"","keywords":"Pooling; Automatic summarization; Computer science; Ranking (information retrieval); Annotation; Artificial intelligence; Principal component analysis; Machine learning; Selection (genetic algorithm)","authors":[{"name":"Philippe Hamel","is_ca":true},{"name":"Simon Lemieux","is_ca":true},{"name":"Yoshua Bengio","is_ca":true},{"name":"Douglas Eck","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05333081680777514,"gpt":0.237022333243612,"spread":0.1836915164358369,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001550621,0.001075032,0.001338394,0.001679398,0.0006113024,0.00114386,0.001427147,0.001162109,0.004617806],"category_scores_gemma":[0.003804239,0.000591294,0.001306283,0.001847412,0.0006678432,0.001740934,0.001477372,0.0009942581,0.001517293],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005219558,"about_ca_system_score_gemma":0.0007067053,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009097008,"about_ca_topic_score_gemma":0.01994569,"domain_scores_codex":[0.9993493,0.0001588346,0.00004097962,0.0001908127,0.0001472203,0.000112774],"domain_scores_gemma":[0.9990734,0.000295204,0.00007167357,0.0002932267,0.0001910295,0.00007546062],"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":[0.0009772413,0.0002259168,0.001293054,0.0003227375,0.0003891524,0.0001949874,0.0001368668,0.02022028,0.1149641,0.01204354,0.0442598,0.8049724],"study_design_scores_gemma":[0.00009223022,0.0002756865,0.004880738,0.00003404483,0.0003236796,0.0002400873,0.00008623859,0.8970202,0.05881012,0.02565733,0.01250687,0.00007285325],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03369353,0.00196193,0.9563365,0.0004831711,0.0003141078,0.0001028015,0.0008577732,0.003867242,0.002383001],"genre_scores_gemma":[0.2944773,0.0009425362,0.6912497,0.0003557286,0.0006198664,0.00017209,0.002882861,0.0006986868,0.008601189],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009097008,"threshold_uncertainty_score":0.0180881,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2166706387","doi":"10.1109/tasl.2006.885921","title":"Audio Signal Feature Extraction and Classification Using Local Discriminant Bases","year":2007,"lang":"en","type":"article","venue":"IEEE Transactions on Audio Speech and Language Processing","topic":"Music and Audio Processing","field":"Computer Science","cited_by":93,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Toronto Metropolitan University; Western University","funders":"","keywords":"Audio signal; Linear discriminant analysis; Flute; Feature extraction; Computer science; Speech recognition; Pattern recognition (psychology); Artificial intelligence; Natural sounds; Classifier (UML); Audio signal processing; Search engine indexing; Linear subspace; Mel-frequency cepstrum; Mathematics; Speech coding; Acoustics","authors":[{"name":"Karthikeyan Umapathy","is_ca":true},{"name":"Sridhar Krishnan","is_ca":true},{"name":"Raveendra K. Rao","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.0230565890800868,"gpt":0.2885249737227967,"spread":0.2654683846427099,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000762506,0.0005307513,0.0007028901,0.001883458,0.0003359104,0.0005962481,0.0005600447,0.0004748384,0.001379556],"category_scores_gemma":[0.002096849,0.0001488552,0.0005739516,0.001307694,0.000254501,0.0005878785,0.0004996062,0.0006890722,0.001323856],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003040637,"about_ca_system_score_gemma":0.0004679652,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002444188,"about_ca_topic_score_gemma":0.001637308,"domain_scores_codex":[0.9994001,0.00008547358,0.00004102131,0.0001181426,0.0002806642,0.00007461441],"domain_scores_gemma":[0.9992393,0.0002170264,0.00006295233,0.00007600346,0.0003637162,0.00004104089],"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":[0.0004770918,0.0002336267,0.003696722,0.0001306238,0.00003841949,0.0001320299,0.00009460106,0.02075599,0.1385955,0.001640601,0.001525358,0.8326793],"study_design_scores_gemma":[0.00006733037,0.0005216377,0.01683959,0.00002958246,0.0000667118,0.0002940979,0.0002135372,0.8916697,0.08322869,0.00233801,0.004666704,0.00006442011],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1244386,0.0004087284,0.8719685,0.0001087254,0.00006408928,0.0001631001,0.0002494038,0.001075487,0.001523384],"genre_scores_gemma":[0.5579677,0.0003435944,0.4368988,0.00005563327,0.00006118825,0.0002284321,0.0009721426,0.00007490953,0.003397591],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.002444188,"threshold_uncertainty_score":0.004859984,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2796517058","doi":"10.3390/app8040606","title":"End-to-End Neural Optical Music Recognition of Monophonic Scores","year":2018,"lang":"en","type":"article","venue":"Applied Sciences","topic":"Music and Audio Processing","field":"Computer Science","cited_by":90,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"Social Sciences and Humanities Research Council of Canada; Ministerio de Economía y Competitividad","keywords":"Connectionism; Computer science; Artificial neural network; Convolutional neural network; Musical notation; Scalability; Notation; Artificial intelligence; Speech recognition; Pattern recognition (psychology); End-to-end principle; Musical; Arithmetic; Mathematics","authors":[{"name":"Jorge Calvo-Zaragoza","is_ca":true},{"name":"David Rizo","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05113765583816901,"gpt":0.2677386336652907,"spread":0.2166009778271217,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0003499844,0.0009888752,0.0005900397,0.000417342,0.0002785629,0.001023671,0.001135348,0.000686525,0.005512889],"category_scores_gemma":[0.001654589,0.0002251051,0.0003717279,0.0004468818,0.0003061024,0.0009112856,0.0007652001,0.0009134458,0.003398841],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004151818,"about_ca_system_score_gemma":0.0006861842,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003276433,"about_ca_topic_score_gemma":0.01034976,"domain_scores_codex":[0.9997161,0.00002730857,0.00001318369,0.0000755201,0.00009807668,0.00006970183],"domain_scores_gemma":[0.9996617,0.00007715332,0.00003133717,0.00006954731,0.0001326427,0.00002760268],"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":[0.0004205117,0.0002602567,0.001479714,0.0001289557,0.00007595101,0.0001706989,0.0000570568,0.04008182,0.05955492,0.001679328,0.006612841,0.889478],"study_design_scores_gemma":[0.00002296748,0.000207192,0.003281629,0.00002241266,0.00003369015,0.0001728511,0.0001115275,0.8999665,0.08753043,0.003727339,0.004898862,0.00002459752],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2394653,0.0007367073,0.7268081,0.0005176759,0.0005177435,0.0002638662,0.001330035,0.01208365,0.01827689],"genre_scores_gemma":[0.6928426,0.0003705854,0.2779459,0.0002633286,0.0001297264,0.0001533039,0.002738069,0.0002665808,0.0252898],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005512889,"threshold_uncertainty_score":0.01844245,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2097705935","doi":"10.1109/tsmcb.2005.862491","title":"Modeling emotional content of music using system identification","year":2006,"lang":"en","type":"article","venue":"IEEE Transactions on Systems Man and Cybernetics Part B (Cybernetics)","topic":"Music and Audio Processing","field":"Computer Science","cited_by":90,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Arousal; Valence (chemistry); Statistic; Ambiguity; Content (measure theory); Identification (biology); Computer science; Emotional valence; Emotion detection; Musical; Speech recognition; Psychology; Artificial intelligence; Pattern recognition (psychology); Cognitive psychology; Mathematics; Emotion recognition; Social psychology; Statistics; Cognition; Art","authors":[{"name":"Mark Korhonen","is_ca":true},{"name":"David A. Clausi","is_ca":true},{"name":"M.E. Jernigan","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05631744724370018,"gpt":0.2370317147277042,"spread":0.180714267484004,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0006379268,0.0004904965,0.000372039,0.0004397819,0.0002265337,0.0006355324,0.0003483883,0.0004371262,0.001041798],"category_scores_gemma":[0.002581041,0.0002241743,0.0004969211,0.0003651534,0.0002457605,0.0008084787,0.0002756415,0.0005016056,0.0002605563],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003713336,"about_ca_system_score_gemma":0.0003083522,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001784376,"about_ca_topic_score_gemma":0.001255724,"domain_scores_codex":[0.9996883,0.0001105234,0.00001623833,0.00007853041,0.00007860347,0.00002789472],"domain_scores_gemma":[0.999437,0.0003723349,0.00005580408,0.0000473259,0.0000783924,0.000009137905],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000154815,0.0001420004,0.01027031,0.000150033,0.000200352,0.0001553856,0.0004813788,0.8104764,0.03154671,0.0115754,0.0005312719,0.134316],"study_design_scores_gemma":[0.000003784344,0.00004049469,0.001149967,0.000003746292,0.00001085666,0.00002182944,0.00001300877,0.9951683,0.001496783,0.001775425,0.0003101514,0.000005665213],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1490647,0.0001721525,0.8475429,0.0001255386,0.0000286931,0.00008370983,0.00008129344,0.0004721023,0.002428951],"genre_scores_gemma":[0.918844,0.0001928625,0.07919711,0.00002773741,0.00002304344,0.0001148416,0.0001411751,0.00002929796,0.001429994],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.001784376,"threshold_uncertainty_score":0.003547966,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1964850609","doi":"10.1109/tifs.2006.885036","title":"Gaussian Mixture Modeling of Short-Time Fourier Transform Features for Audio Fingerprinting","year":2006,"lang":"en","type":"article","venue":"IEEE Transactions on Information Forensics and Security","topic":"Music and Audio Processing","field":"Computer Science","cited_by":87,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Toronto Metropolitan University","funders":"","keywords":"Computer science; Pattern recognition (psychology); Mixture model; Centroid; Artificial intelligence; Fingerprint (computing); Speech recognition; Cepstrum; Audio signal; Robustness (evolution); Entropy (arrow of time); Word error rate; Speech coding","authors":[{"name":"A. Ramalingam","is_ca":true},{"name":"Sridhar Krishnan","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.007905369170801714,"gpt":0.2148253334217568,"spread":0.2069199642509551,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001509195,0.0008528316,0.0008266639,0.00110239,0.0002641215,0.000666954,0.0009155902,0.0008147369,0.001585073],"category_scores_gemma":[0.004324282,0.0003186052,0.001051579,0.001080315,0.0003702177,0.001288451,0.0004111257,0.001032125,0.001122328],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006194031,"about_ca_system_score_gemma":0.0004601717,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006555375,"about_ca_topic_score_gemma":0.004517888,"domain_scores_codex":[0.9993631,0.0001596195,0.00002906975,0.0001196072,0.0002561816,0.0000725035],"domain_scores_gemma":[0.9988011,0.0006953752,0.00009008693,0.0001480134,0.0002375973,0.00002770725],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0004640244,0.0001547858,0.003204299,0.0001264464,0.0001414686,0.0001650618,0.0001222716,0.6293078,0.03069236,0.009682002,0.002616838,0.3233226],"study_design_scores_gemma":[0.000002843613,0.00002111992,0.0006880402,0.000003834426,0.00001193231,0.00003248841,0.000006535088,0.9944491,0.003035783,0.001154065,0.0005814368,0.00001287766],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02112811,0.0005287189,0.976252,0.00009668143,0.00007906418,0.00004280648,0.0001278405,0.00105972,0.000685191],"genre_scores_gemma":[0.6292579,0.001103005,0.3639767,0.00009554465,0.0001096069,0.0001244791,0.0007851794,0.0002912479,0.00425633],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.006555375,"threshold_uncertainty_score":0.01303446,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W1688667826","doi":"10.1073/pnas.1510724112","title":"Cross-cultural convergence of musical features","year":2015,"lang":"en","type":"letter","venue":"Proceedings of the National Academy of Sciences","topic":"Music and Audio Processing","field":"Computer Science","cited_by":87,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Toronto","funders":"","keywords":"Musical; Convergence (economics); Computer science; Computational biology; Speech recognition; Biological system; Biology; Art; Economics; Literature","authors":[{"name":"Sandra E. Trehub","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06741783976302491,"gpt":0.3420739352726175,"spread":0.2746560955095926,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003664309,0.0001200766,0.0002729675,0.0005698671,0.001186312,0.001848487,0.0004923907,0.002608538,0.008677134],"category_scores_gemma":[0.01227307,0.0001243755,0.0001720396,0.0004595919,0.002485702,0.001236315,0.002027803,0.003971246,0.001254511],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001109467,"about_ca_system_score_gemma":0.0004183807,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004117437,"about_ca_topic_score_gemma":0.008919423,"domain_scores_codex":[0.9979542,0.0008399341,0.000113478,0.0003729208,0.0004381732,0.0002812078],"domain_scores_gemma":[0.9934818,0.003449688,0.0004257135,0.001118202,0.001096014,0.0004284951],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.001114969,0.0001797762,0.1313537,0.000216389,0.0003157243,0.009240874,0.06152712,0.0002449641,0.009948757,0.122275,0.1190538,0.5445289],"study_design_scores_gemma":[0.0001784505,0.0004333128,0.3781522,0.0009433377,0.0001635593,0.01658162,0.06792586,0.0009731828,0.005928434,0.1384501,0.3901144,0.0001555391],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.4617797,0.00539905,0.002342238,0.3793079,0.002078527,0.00003076931,0.0002879165,0.00003558752,0.1487383],"genre_scores_gemma":[0.9550867,0.001230902,0.0002592166,0.03478368,0.0009243409,0.00001980699,0.00006060522,0.00001804133,0.007616549],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.008677134,"threshold_uncertainty_score":0.02902788,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3049459827","doi":"10.1109/cjece.2020.2970144","title":"Neural Network Music Genre Classification","year":2020,"lang":"en","type":"article","venue":"Canadian Journal of Electrical and Computer Engineering","topic":"Music and Audio Processing","field":"Computer Science","cited_by":84,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":true,"about_ca":false},"ca_institutions":"University of Regina","funders":"","keywords":"Spectrogram; Computer science; Artificial neural network; Musical; Artificial intelligence; Speech recognition; Machine learning; Art; Visual arts","authors":[{"name":"Nikki Pelchat","is_ca":true},{"name":"Craig Gelowitz","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02022962826200411,"gpt":0.1746245635508428,"spread":0.1543949352888387,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005160421,0.0007294317,0.0005845682,0.001867172,0.0003884631,0.001037833,0.0007770568,0.0007211441,0.005574543],"category_scores_gemma":[0.002074215,0.0002046038,0.0005536809,0.001662302,0.0001946993,0.0008390673,0.0004645178,0.0008333884,0.003134408],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005992625,"about_ca_system_score_gemma":0.0003378221,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007384907,"about_ca_topic_score_gemma":0.006742534,"domain_scores_codex":[0.9997229,0.00003925214,0.00002278748,0.00007137532,0.00009363962,0.00005007046],"domain_scores_gemma":[0.9996891,0.00008084416,0.00003013853,0.00002799362,0.000152319,0.00001959992],"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":[0.0002643158,0.0001819618,0.006366007,0.0001659296,0.0001235024,0.0001321843,0.00004580977,0.06027117,0.0105343,0.001682285,0.01184247,0.9083902],"study_design_scores_gemma":[0.00001706677,0.00007318659,0.005746968,0.00006998722,0.00006047025,0.000134084,0.00007789981,0.9788745,0.004388554,0.002823026,0.007711152,0.00002302611],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2924469,0.02119361,0.5786629,0.002501896,0.002890092,0.0004763785,0.00420845,0.00812786,0.08949202],"genre_scores_gemma":[0.827911,0.005816743,0.1213696,0.0005782432,0.00074167,0.0001638093,0.004511855,0.0002047556,0.03870231],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007384907,"threshold_uncertainty_score":0.01864874,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3109392859","doi":"10.3989/loquens.2019.065","title":"Sets, rules and natural classes: [ ] vs. { }","year":2019,"lang":"en","type":"article","venue":"Loquens","topic":"Music and Audio Processing","field":"Computer Science","cited_by":83,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Concordia University","funders":"","keywords":"Unification; Computer science; Feature (linguistics); Set (abstract data type); Subtraction; Theoretical computer science; Natural (archaeology); Singleton; Artificial intelligence; Mathematics; Algorithm; Arithmetic; Programming language; Linguistics","authors":[{"name":"Alan Bale","is_ca":true},{"name":"Charles Reiss","is_ca":true},{"name":"David Ta-Chun Shen","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.007942647612564515,"gpt":0.2297369402026419,"spread":0.2217942925900774,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002280976,0.0004358439,0.0005216218,0.001238123,0.002061687,0.004820894,0.001223799,0.001667725,0.004798342],"category_scores_gemma":[0.00335796,0.0004542036,0.0008676457,0.001049509,0.01120921,0.006348311,0.002048975,0.002554451,0.0007639225],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001817272,"about_ca_system_score_gemma":0.0007463152,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004001318,"about_ca_topic_score_gemma":0.002804667,"domain_scores_codex":[0.9984275,0.0006926006,0.0001219409,0.0003289967,0.0002567032,0.0001721785],"domain_scores_gemma":[0.9985805,0.0007833033,0.0001318204,0.000294912,0.000120896,0.00008860812],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00001579789,0.000004034266,0.0001063775,0.0000186558,0.000002851777,0.00002799532,0.0003720497,0.000303177,0.0004261867,0.9918146,0.0004059979,0.006502242],"study_design_scores_gemma":[0.00001724564,0.00002637029,0.0002699895,0.00003708614,0.00001587283,0.0001474103,0.0003453741,0.004892844,0.001806106,0.9553391,0.03707571,0.00002688279],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"other","genre_scores_codex":[0.07591169,0.004158215,0.6902497,0.005718186,0.0008957886,0.0001444778,0.0002797642,0.0007940767,0.221848],"genre_scores_gemma":[0.7931144,0.0007436954,0.1927679,0.001180024,0.0002305343,0.0002474888,0.0001891075,0.0002864666,0.01124044],"genre_candidate":"other","genre_consensus":null,"teacher_disagreement_score":0.004820894,"threshold_uncertainty_score":0.01605207,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2753868141","doi":"10.48550/arxiv.1903.07227","title":"Counterpoint By Convolution.","year":2017,"lang":"en","type":"article","venue":"arXiv (Cornell University)","topic":"Music and Audio Processing","field":"Computer Science","cited_by":80,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal","funders":"","keywords":"Gibbs sampling; Computer science; Sampling (signal processing); Convolution (computer science); Convolutional neural network; Artificial intelligence; Counterpoint; Set (abstract data type); Algorithm; Artificial neural network; Filter (signal processing); Computer vision; Programming language","authors":[{"name":"Cheng-Zhi Anna Huang","is_ca":false},{"name":"Tim Cooijmans","is_ca":false},{"name":"Adam P. Roberts","is_ca":false},{"name":"Aaron Courville","is_ca":true},{"name":"Douglas Eck","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.05705792341172985,"gpt":0.1739792359229458,"spread":0.116921312511216,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001904228,0.001138184,0.001200878,0.0009859867,0.001420992,0.002772509,0.002681011,0.003029308,0.0466064],"category_scores_gemma":[0.0117867,0.0006267611,0.001418879,0.0008678756,0.002802663,0.008240438,0.004838735,0.003270218,0.009044116],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00118066,"about_ca_system_score_gemma":0.001296572,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001637756,"about_ca_topic_score_gemma":0.002169033,"domain_scores_codex":[0.9979226,0.0004390461,0.0001115987,0.0007317968,0.0004772632,0.0003177203],"domain_scores_gemma":[0.9970777,0.001117076,0.0002284803,0.001065636,0.0003370947,0.0001740816],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0005028688,0.00007688561,0.00134563,0.0002928507,0.0001138186,0.0005238628,0.0002455928,0.01887483,0.003883836,0.8216951,0.03869477,0.1137499],"study_design_scores_gemma":[0.00007223361,0.00005249997,0.0002903686,0.000075731,0.0000615269,0.0008572825,0.00008435868,0.1702602,0.005626262,0.7672616,0.05532016,0.00003783275],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01507076,0.002136323,0.9028021,0.00358519,0.002050288,0.0001569209,0.000590191,0.004182955,0.06942526],"genre_scores_gemma":[0.589456,0.00156401,0.3362679,0.00289553,0.001154707,0.0004556376,0.001117244,0.001669779,0.06541921],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.0466064,"threshold_uncertainty_score":0.1559139,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2152156982","doi":"10.1109/ism.2009.123","title":"Music Emotion Identification from Lyrics","year":2009,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":79,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Ottawa","funders":"","keywords":"Lyrics; Computer science; Music information retrieval; Identification (biology); Emotion classification; Musical; Emotion detection; Natural language processing; Classifier (UML); Emotion recognition; Artificial intelligence; Art; Literature","authors":[{"name":"Yang Dan","is_ca":true},{"name":"Won‐Sook Lee","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02320938488149102,"gpt":0.2355040165623796,"spread":0.2122946316808886,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004328375,0.0006498502,0.0005091623,0.003783518,0.0003708535,0.001218208,0.0003885127,0.0007402985,0.004382978],"category_scores_gemma":[0.003559041,0.0001293936,0.0004499346,0.001652058,0.0002392011,0.0008827616,0.0008020871,0.0004650288,0.004693614],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002692403,"about_ca_system_score_gemma":0.0002005933,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008171219,"about_ca_topic_score_gemma":0.001553481,"domain_scores_codex":[0.9994982,0.0001021925,0.00004837139,0.0001208178,0.0001579432,0.0000723289],"domain_scores_gemma":[0.9985867,0.0003085749,0.0001622503,0.0001662846,0.0006839391,0.00009240401],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.001660134,0.0001962466,0.03275505,0.001062231,0.0001475878,0.0008934286,0.001308204,0.002960866,0.1911532,0.001925745,0.01849756,0.7474398],"study_design_scores_gemma":[0.0001874273,0.001068959,0.4738349,0.0003432978,0.0002867969,0.003180383,0.005844732,0.2882017,0.1416025,0.004905409,0.08033121,0.0002125777],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.7524808,0.003985232,0.1884679,0.0006304632,0.0003817897,0.0008468392,0.01357555,0.004891627,0.03473972],"genre_scores_gemma":[0.8469277,0.001580142,0.1179626,0.0001245902,0.0003970834,0.0002960181,0.02047673,0.0002947181,0.01194042],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.004382978,"threshold_uncertainty_score":0.0146625,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W192383697","doi":"10.5281/zenodo.7432188","title":"ENST-Drums: an extensive audio-visual database for drum signals processing","year":2006,"lang":"en","type":"dataset","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":77,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true},"ca_institutions":"","funders":"","keywords":"Drum; Audio visual; Computer science; Database; Computer graphics (images); Speech recognition; Multimedia; Geography; Archaeology","authors":[{"name":"Olivier Gillet","is_ca":false},{"name":"Gaël Richard","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02936108337964647,"gpt":0.3172990123547395,"spread":0.287937928975093,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001438406,0.003330701,0.00211203,0.005605123,0.0007546469,0.002450256,0.003769224,0.002401889,0.07530822],"category_scores_gemma":[0.005499634,0.0007659606,0.001275012,0.004759989,0.0004046141,0.003000141,0.003479342,0.001282786,0.1154539],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000840756,"about_ca_system_score_gemma":0.001654658,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008356727,"about_ca_topic_score_gemma":0.0100501,"domain_scores_codex":[0.9975408,0.0003270348,0.0004404165,0.0005382303,0.0009163621,0.0002370957],"domain_scores_gemma":[0.9973925,0.0004275294,0.0001946749,0.0006579915,0.001084178,0.0002430293],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.001045234,0.0001252571,0.0009099817,0.002415229,0.00008958733,0.0003512902,0.000199668,0.001471892,0.01264123,0.001421498,0.9082284,0.07110079],"study_design_scores_gemma":[0.0005013623,0.0002324223,0.01052397,0.0004575916,0.0001272905,0.0007287408,0.0003611782,0.009739895,0.01737419,0.004391589,0.9552823,0.0002794462],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"dataset","genre_gemma":"dataset","genre_scores_codex":[0.003029477,0.001178076,0.02638013,0.0001423301,0.0002853224,0.0002950122,0.9081199,0.04967403,0.01089576],"genre_scores_gemma":[0.005646039,0.0003338306,0.01441081,0.00008033793,0.000053264,0.0003755231,0.9711173,0.003225253,0.004757653],"genre_candidate":"dataset","genre_consensus":"dataset","teacher_disagreement_score":0.07530822,"threshold_uncertainty_score":0.2519311,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2186385212","doi":"10.31234/osf.io/s9ryg","title":"CantoCore: A new cross-cultural song classification scheme","year":2020,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":77,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"McMaster University","funders":"","keywords":"Classification scheme; Scheme (mathematics); Musical; Problem of universals; Computer science; Natural language processing; Reliability (semiconductor); Cross-cultural; Artificial intelligence; Linguistics; Machine learning; Literature; Mathematics; Art; Sociology; Anthropology; Philosophy","authors":[{"name":"Patrick E. Savage","is_ca":true},{"name":"Emily Merritt","is_ca":true},{"name":"Tom Rzeszutek","is_ca":true},{"name":"Steven Brown","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1010133824232493,"gpt":0.3313369098939356,"spread":0.2303235274706864,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00483115,0.0008116544,0.0006141894,0.01210143,0.001682882,0.002633922,0.001379747,0.0007817987,0.005734938],"category_scores_gemma":[0.02384307,0.0002725254,0.0007339554,0.005422362,0.001906067,0.003627025,0.003807848,0.001107056,0.001994127],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00128196,"about_ca_system_score_gemma":0.002030649,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006891465,"about_ca_topic_score_gemma":0.007072165,"domain_scores_codex":[0.9960841,0.0008886105,0.0008203113,0.0005969243,0.001278877,0.0003311851],"domain_scores_gemma":[0.9856195,0.003225085,0.001581226,0.002517829,0.006270699,0.0007856121],"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":[0.001318007,0.000203562,0.1028895,0.0006044027,0.0001686715,0.0002222519,0.003130254,0.003632676,0.016613,0.04429179,0.01756264,0.8093633],"study_design_scores_gemma":[0.0003901482,0.001707918,0.3448727,0.0008719911,0.0003348422,0.002008585,0.008986477,0.2120478,0.02828302,0.1179337,0.2815321,0.001030741],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.3573934,0.001193571,0.5492743,0.001278696,0.001125779,0.002284917,0.01177562,0.005520676,0.07015309],"genre_scores_gemma":[0.5985709,0.0002735471,0.3801813,0.0002908469,0.0002045546,0.001293058,0.009967365,0.0004665568,0.008751812],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01210143,"threshold_uncertainty_score":0.02554989,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2591984255","doi":"10.17863/cam.21343","title":"Sequence Tutor: Conservative Fine-Tuning of Sequence Generation Models with KL-control","year":2017,"lang":"en","type":"article","venue":"Apollo (University of Cambridge)","topic":"Music and Audio Processing","field":"Computer Science","cited_by":76,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal; Google (Canada)","funders":"Engineering and Physical Sciences Research Council; Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Recurrent neural network; Sequence (biology); Reinforcement learning; Security token; Artificial intelligence; Machine learning; Artificial neural network","authors":[{"name":"Natasha Jaques","is_ca":false},{"name":"Shixiang Gu","is_ca":false},{"name":"José Miguel Hernández-Lobato","is_ca":false},{"name":"Richard E. Turner","is_ca":false},{"name":"Douglas Eck","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06863170625603472,"gpt":0.239950264465594,"spread":0.1713185582095593,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002813745,0.001104185,0.0009655235,0.0005245131,0.0004309006,0.00091765,0.002022626,0.001137228,0.003388411],"category_scores_gemma":[0.009267523,0.0005511149,0.0005067873,0.0002765758,0.001269797,0.001453674,0.001970183,0.001970778,0.000722109],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008912724,"about_ca_system_score_gemma":0.001236474,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002760936,"about_ca_topic_score_gemma":0.003236106,"domain_scores_codex":[0.9989681,0.0003393711,0.00005196088,0.0002269354,0.0003183266,0.00009527774],"domain_scores_gemma":[0.9969459,0.001947869,0.0002724176,0.0003518652,0.0003441104,0.0001379097],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001115335,0.00009353199,0.0005315164,0.00006668696,0.00003161135,0.00006094513,0.0000865272,0.9096749,0.004675353,0.01015799,0.0009673316,0.07354204],"study_design_scores_gemma":[0.000007864462,0.00001981654,0.00002226867,0.000002851407,0.000001879105,0.000005986083,0.000002153412,0.9975029,0.0006366239,0.001604012,0.0001908908,0.000002815917],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01092962,0.0001133709,0.9860982,0.00008431326,0.0000390987,0.00004471779,0.00001719666,0.001056753,0.001616814],"genre_scores_gemma":[0.7473419,0.0001040898,0.2474133,0.0003119206,0.00008203053,0.0002400753,0.0001350266,0.0005788886,0.003792728],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003388411,"threshold_uncertainty_score":0.01488066,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2138960858","doi":"","title":"Modeling Deep Temporal Dependencies with Recurrent Grammar Cells","year":2014,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":76,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université de Montréal","funders":"","keywords":"Computer science; Recurrent neural network; Frame (networking); Autoencoder; Variety (cybernetics); Artificial intelligence; Transformation (genetics); Syntax; Grammar; Pyramid (geometry); Artificial neural network; Pattern recognition (psychology); Algorithm; Natural language processing; Mathematics; Linguistics","authors":[{"name":"Vincent Michalski","is_ca":false},{"name":"Roland Memisevic","is_ca":true},{"name":"Kishore Konda","is_ca":false}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01618343484872841,"gpt":0.212876189027172,"spread":0.1966927541784436,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005079616,0.0006826281,0.0006545831,0.0004077054,0.0002508166,0.0009047277,0.001449866,0.001036648,0.002099767],"category_scores_gemma":[0.001937958,0.0006042945,0.0008161869,0.0005027815,0.0006508032,0.001510428,0.0007985774,0.001499297,0.0005937175],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007801968,"about_ca_system_score_gemma":0.0009705353,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.009496185,"about_ca_topic_score_gemma":0.01739755,"domain_scores_codex":[0.9997756,0.00005130237,0.00001099096,0.00007576483,0.00004067717,0.00004579938],"domain_scores_gemma":[0.9995226,0.0002359795,0.00006627238,0.00006643868,0.00007443462,0.00003435557],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0000804076,0.00005677742,0.001564741,0.0000442329,0.00006595178,0.0002594659,0.00014999,0.8903539,0.0120995,0.04062986,0.001729758,0.05296547],"study_design_scores_gemma":[0.00000359597,0.000009566727,0.000091347,0.000002121978,0.000005623377,0.00001139043,0.000004013216,0.9882811,0.0007245061,0.01059917,0.0002641047,0.000003504639],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07200295,0.0002492791,0.9232999,0.0003486847,0.00007808535,0.00002562493,0.0003933563,0.001551531,0.002050575],"genre_scores_gemma":[0.8645507,0.000377641,0.1281806,0.0001848735,0.00005553836,0.00009259187,0.000721031,0.000234985,0.00560218],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.009496185,"threshold_uncertainty_score":0.0188818,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2328848613","doi":"10.1080/09298215.2015.1132737","title":"A Comparison of Approaches to Timbre Descriptors in Music Information Retrieval and Music Psychology","year":2016,"lang":"en","type":"article","venue":"Journal of New Music Research","topic":"Music and Audio Processing","field":"Computer Science","cited_by":75,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false},"ca_institutions":"McGill University","funders":"Fonds de recherche du Québec – Nature et technologies; Natural Sciences and Engineering Research Council of Canada; Canada Research Chairs","keywords":"Timbre; Music information retrieval; Information retrieval; Computer science; Speech recognition; Psychology; Musical; Visual arts; Art","authors":[{"name":"Kai Siedenburg","is_ca":true},{"name":"Ichiro Fujinaga","is_ca":true},{"name":"Stephen McAdams","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.5352193071220087,"gpt":0.435350893729382,"spread":0.0998684133926267,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01302623,0.0008558048,0.00128982,0.01362644,0.001294147,0.01306683,0.002671224,0.002758388,0.005499642],"category_scores_gemma":[0.02918267,0.0005996338,0.001620394,0.01164887,0.007668719,0.01280552,0.004029149,0.002682611,0.00186122],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004047089,"about_ca_system_score_gemma":0.002498561,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002263516,"about_ca_topic_score_gemma":0.001898421,"domain_scores_codex":[0.9910473,0.005436209,0.0006705508,0.0007544427,0.00180635,0.0002850746],"domain_scores_gemma":[0.9810932,0.01378374,0.0008978267,0.001905794,0.001914074,0.0004052899],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0001714414,0.0001124122,0.00171502,0.001232716,0.0001368137,0.00007793164,0.00508422,0.002559948,0.00218216,0.6277363,0.002890628,0.3561005],"study_design_scores_gemma":[0.00008790136,0.0002233691,0.01125239,0.001249656,0.0001687754,0.0007184821,0.007572103,0.03540574,0.002637916,0.8559616,0.08447333,0.0002487824],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01698622,0.03714095,0.8794704,0.01080339,0.0004085353,0.0002866259,0.0001962909,0.0004381716,0.05426942],"genre_scores_gemma":[0.4466791,0.02626429,0.5103186,0.002668516,0.001527464,0.0008488729,0.0005940667,0.0003992001,0.01069987],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01362644,"threshold_uncertainty_score":0.06889015,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2078120539","doi":"10.1145/2009916.2010011","title":"Enhancing multi-label music genre classification through ensemble techniques","year":2011,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":75,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Lethbridge","funders":"","keywords":"Computer science; Multi-label classification; Ensemble learning; Set (abstract data type); Artificial intelligence; Music information retrieval; Task (project management); Machine learning; Field (mathematics); Statistical classification; Empirical research; Natural language processing; Information retrieval; Musical; Mathematics; Engineering","authors":[{"name":"Chris Sanden","is_ca":true},{"name":"John Z. Zhang","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1637955893780131,"gpt":0.3016030120560653,"spread":0.1378074226780522,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003557374,0.001242914,0.001735144,0.003015735,0.0009309376,0.001301437,0.001165942,0.001419056,0.00139494],"category_scores_gemma":[0.007757585,0.0003444479,0.001256811,0.002393077,0.0003993035,0.002692389,0.001236309,0.002350183,0.001181074],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003620442,"about_ca_system_score_gemma":0.0005034083,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001742066,"about_ca_topic_score_gemma":0.003214135,"domain_scores_codex":[0.9984323,0.0005063928,0.00008413737,0.000296298,0.0005447209,0.000136081],"domain_scores_gemma":[0.9950438,0.00215988,0.0003386355,0.0007068267,0.00157479,0.0001761339],"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":[0.0002604118,0.0003760226,0.004724596,0.0001065224,0.0002897017,0.00009357461,0.000302134,0.04458014,0.02455104,0.002152948,0.005892552,0.9166703],"study_design_scores_gemma":[0.00002710904,0.0002648865,0.00317123,0.00003164433,0.0001877168,0.0002182159,0.0001927448,0.9712362,0.01314163,0.00698339,0.004492865,0.00005246115],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.04501374,0.001333359,0.9484019,0.000437933,0.000277439,0.0000938518,0.0001256836,0.001499483,0.00281658],"genre_scores_gemma":[0.4313158,0.001025097,0.5609714,0.0004295286,0.0006735522,0.0001648639,0.0009726801,0.0002309792,0.004216031],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003557374,"threshold_uncertainty_score":0.01881337,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2107495852","doi":"10.1109/icme.2006.262954","title":"Mixed Type Audio Classification with Support Vector Machine","year":2006,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":75,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Computer science; Audio mining; Support vector machine; Speech recognition; Naive Bayes classifier; Feature extraction; Classifier (UML); Artificial intelligence; Mel-frequency cepstrum; Audio signal processing; Audio signal; Pattern recognition (psychology); Speech coding; Acoustic model; Speech processing","authors":[{"name":"Lei Chen","is_ca":false},{"name":"Şule Gündüz Öğüdücü","is_ca":false},{"name":"M. TAMER ÖZSU","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.02580508025301679,"gpt":0.2253374570773634,"spread":0.1995323768243466,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001678483,0.001209775,0.001166341,0.003166646,0.0004176377,0.001095871,0.001203077,0.001236652,0.003331589],"category_scores_gemma":[0.006376869,0.0004073537,0.0008422004,0.001649155,0.0002845101,0.001603239,0.0006666988,0.001048734,0.002339058],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003905892,"about_ca_system_score_gemma":0.0004124054,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001700264,"about_ca_topic_score_gemma":0.001412926,"domain_scores_codex":[0.9985085,0.0003722365,0.0001735204,0.0002955225,0.0005327615,0.0001174269],"domain_scores_gemma":[0.9967769,0.001509726,0.0002557832,0.0003049494,0.001055057,0.00009761949],"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":[0.0004345473,0.0002013843,0.002514766,0.0001588257,0.0001298587,0.0001079768,0.00004275103,0.04168754,0.01290836,0.00088646,0.003745663,0.9371819],"study_design_scores_gemma":[0.00002113855,0.00008721078,0.0009551791,0.00001334786,0.00002318843,0.00007935136,0.0000274512,0.9900135,0.006067667,0.001764067,0.0009299609,0.00001792798],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.03058549,0.0005043888,0.9625099,0.0001431143,0.0001554939,0.0001094529,0.0003182521,0.00474912,0.0009247648],"genre_scores_gemma":[0.3721893,0.0003077971,0.6228625,0.0001217996,0.0002258933,0.0002934072,0.001070744,0.0001361936,0.002792278],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003331589,"threshold_uncertainty_score":0.01114529,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W3139654456","doi":"10.1109/infvis.2004.60","title":"Steerable, Progressive Multidimensional Scaling","year":2005,"lang":"en","type":"article","venue":"IEEE Symposium on Information Visualization","topic":"Music and Audio Processing","field":"Computer Science","cited_by":71,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of British Columbia","funders":"","keywords":"Scaling; Computer science; Multidimensional scaling; Mathematics; Machine learning; Geometry","authors":[{"name":"Matt Williams","is_ca":true},{"name":"Tamara Munzner","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.01085608954989996,"gpt":0.2703813822750954,"spread":0.2595252927251955,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001267249,0.001148759,0.0009536775,0.001451278,0.0009207824,0.002532208,0.001815859,0.0009667802,0.004951177],"category_scores_gemma":[0.006594022,0.0008060583,0.001127402,0.001804729,0.001002358,0.002977285,0.00380136,0.00126216,0.001919652],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005658239,"about_ca_system_score_gemma":0.001065323,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002333008,"about_ca_topic_score_gemma":0.004776532,"domain_scores_codex":[0.9988393,0.000229775,0.00006733213,0.0002270534,0.0005707142,0.00006590789],"domain_scores_gemma":[0.9981055,0.0005624646,0.0001576996,0.0006505051,0.000408897,0.0001149232],"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":[0.0004888918,0.0002093212,0.002536571,0.0005921361,0.0001609768,0.000393967,0.001917326,0.2439664,0.06774238,0.08381119,0.02836801,0.5698128],"study_design_scores_gemma":[0.00008334355,0.0001565096,0.0007313066,0.00005135342,0.00002201067,0.0003451059,0.0003291694,0.8729034,0.02764577,0.0584252,0.03921954,0.00008727067],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.007034051,0.0001109684,0.9879743,0.000124312,0.00003044836,0.00006648093,0.0001539797,0.003389106,0.001116467],"genre_scores_gemma":[0.08919757,0.0002954587,0.9065892,0.00009620155,0.00002375689,0.0002407174,0.0004743364,0.0007559993,0.002326723],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004951177,"threshold_uncertainty_score":0.01656336,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2538750502","doi":"10.1109/iembs.2004.1403077","title":"Respiratory sounds classification using cepstral analysis and Gaussian mixture models","year":2005,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":66,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"Université du Québec à Rimouski","funders":"","keywords":"Mel-frequency cepstrum; Pattern recognition (psychology); Speech recognition; Vector quantization; Mixture model; Computer science; Artificial intelligence; Multilayer perceptron; Feature vector; Perceptron; Feature extraction; Cepstrum; Hidden Markov model; Artificial neural network; Support vector machine; Gaussian","authors":[{"name":"Mohammed Bahoura","is_ca":true},{"name":"Céline Pelletier","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.06297010440452305,"gpt":0.2887242472730468,"spread":0.2257541428685237,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009174017,0.0005495074,0.0007131189,0.001604671,0.0002089785,0.0008223334,0.0004294135,0.000798666,0.0008599795],"category_scores_gemma":[0.002346834,0.000216782,0.0006507548,0.0008498527,0.0002310848,0.0008259136,0.0002896377,0.000527909,0.0008880435],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003429917,"about_ca_system_score_gemma":0.0003431189,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00452984,"about_ca_topic_score_gemma":0.003290864,"domain_scores_codex":[0.9994015,0.0001553698,0.00004107602,0.0001138304,0.0002233779,0.00006481125],"domain_scores_gemma":[0.9994323,0.0002055056,0.00003903101,0.000047904,0.0002544523,0.00002079287],"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":[0.0004861956,0.0001634967,0.005898697,0.0002084172,0.0001458552,0.0002364449,0.00009588368,0.0710857,0.06113385,0.003758876,0.004998457,0.851788],"study_design_scores_gemma":[0.00003013116,0.0001230526,0.009201995,0.0000299958,0.00008253887,0.0002352713,0.00004317229,0.9721238,0.01329467,0.001729585,0.003055023,0.00005072542],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04947855,0.001955468,0.9448677,0.0002018393,0.0001560773,0.00006574571,0.0002230968,0.001382285,0.001669319],"genre_scores_gemma":[0.6750132,0.001771621,0.31797,0.0001046958,0.0001922565,0.00009497926,0.0008426023,0.00009555036,0.003915187],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00452984,"threshold_uncertainty_score":0.009006917,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null},{"id":"W2132189140","doi":"10.18061/1811/48548","title":"Using Automated Rhyme Detection to Characterize Rhyming Style in Rap Music","year":2010,"lang":"en","type":"article","venue":"Empirical Musicology Review","topic":"Music and Audio Processing","field":"Computer Science","cited_by":65,"is_retracted":false,"has_abstract":true,"routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false},"ca_institutions":"University of Waterloo","funders":"","keywords":"Rhyme; Style (visual arts); Psychology; Speech recognition; Communication; Art; Linguistics; Computer science; Literature; Philosophy; Poetry","authors":[{"name":"Hussein Hirjee","is_ca":true},{"name":"Daniel G. Brown","is_ca":true}],"retraction":null,"screen_n_in":null,"score":{"opus":0.1135838939556501,"gpt":0.3659130687035156,"spread":0.2523291747478655,"validation_status":"score_only:v0-immature-baseline"},"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001628101,0.0005507945,0.000591143,0.003693404,0.0002712878,0.001032978,0.000658229,0.0006202199,0.001644929],"category_scores_gemma":[0.006434313,0.0002047779,0.0003724699,0.001823531,0.0003183948,0.0009347756,0.0005095097,0.0004086862,0.00174915],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002442291,"about_ca_system_score_gemma":0.0003376455,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001742601,"about_ca_topic_score_gemma":0.002526059,"domain_scores_codex":[0.9986913,0.0004823778,0.00009891762,0.0002967986,0.0003515593,0.00007912059],"domain_scores_gemma":[0.9952796,0.002151205,0.0007364007,0.0005643357,0.001146154,0.0001223819],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.0005771428,0.000211761,0.0595563,0.0006106275,0.0002292445,0.000175928,0.0004274,0.008455455,0.08543882,0.001022092,0.002611347,0.8406839],"study_design_scores_gemma":[0.0001337454,0.0006708523,0.2300546,0.0001235324,0.0002351872,0.001190318,0.0005639126,0.6724688,0.08525641,0.003389683,0.00573652,0.0001764293],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5031525,0.001505495,0.4833149,0.0001577748,0.00009769429,0.000245599,0.001364599,0.003627008,0.006534367],"genre_scores_gemma":[0.8223419,0.0004552996,0.1735659,0.00003727143,0.00008051913,0.0001307513,0.001112864,0.0001120524,0.002163581],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.003693404,"threshold_uncertainty_score":0.008610308,"prediction_status":"machine_predicted_unvalidated"},"labels":[],"label_agreement":null}]}