{"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,"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","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0002128643,0.00009570768,0.00009704208,0.00009937483,0.0001796262,0.0001600126,0.0001195498,0.00004147024,0.00002065961],"category_scores_gemma":[0.00001457221,0.00007188765,0.00001303106,0.0005636938,0.00001697599,0.0003849753,0.0001024597,0.00008545841,0.00003218816],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00001790899,"about_ca_system_score_gemma":0.00002472211,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000007512777,"about_ca_topic_score_gemma":0.00005019678,"domain_scores_codex":[0.9991775,0.00002332112,0.00007727752,0.0003047683,0.0001990515,0.0002180747],"domain_scores_gemma":[0.9996723,0.0000378047,0.0000450384,0.0001219782,0.00004410498,0.00007878541],"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":[0.0001145215,0.00008736847,0.009606932,0.0001560318,0.00006616784,0.000108142,0.2051406,0.0008068438,0.0810886,0.006107971,0.02789843,0.6688184],"study_design_scores_gemma":[0.00530858,0.0007514471,0.2856543,0.0008579382,0.000061183,0.0001023615,0.03470401,0.6352397,0.02938636,0.001401277,0.005160846,0.001372013],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3496008,0.00003837464,0.6457483,0.002010686,0.00007281828,0.0001329394,0.000001903033,0.0002616458,0.002132552],"genre_scores_gemma":[0.943721,0.000007769921,0.05212638,0.001479199,0.00007328064,0.00001523879,0.00001112398,0.000008722618,0.002557296],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.6674464,"threshold_uncertainty_score":0.2931494,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.00899152659473625,"score_gpt":0.2488691384566329,"score_spread":0.2398776118618966,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}