{"id":"W3209976096","doi":"10.1109/icassp43922.2022.9746832","title":"Pseudo-Labeling for Massively Multilingual Speech Recognition","year":2022,"lang":"en","type":"article","venue":"ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Computer science; Natural language processing; Speech recognition; Artificial intelligence; Language model","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":["metaepi_narrow","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001281981,0.0005043268,0.0004952953,0.0006453044,0.001170739,0.0009734277,0.001441113,0.0001717563,0.001835705],"category_scores_gemma":[0.0004871654,0.0005453718,0.0002125792,0.0005084452,0.0001591545,0.00068632,0.0004445306,0.0009769145,0.00008054147],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003592802,"about_ca_system_score_gemma":0.0005093205,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003252007,"about_ca_topic_score_gemma":0.000008942017,"domain_scores_codex":[0.9952542,0.0002226744,0.0008510088,0.001318017,0.001658459,0.0006956261],"domain_scores_gemma":[0.9970943,0.0006030964,0.0005837607,0.0003671684,0.001038551,0.0003130762],"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":[0.000299814,0.0004823565,0.00003833366,0.0000722452,0.0001065032,0.0002795746,0.0003380717,0.0001657732,0.08278895,0.001034004,0.002721942,0.9116724],"study_design_scores_gemma":[0.002081087,0.0007859121,0.00007947147,0.0001943046,0.00009493511,0.0005887453,0.001638339,0.9364178,0.02974969,0.02294688,0.004309768,0.001113107],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2025159,0.0004637695,0.7569371,0.01083618,0.006903965,0.002252006,0.002486869,0.001177175,0.01642704],"genre_scores_gemma":[0.8362169,0.0001937094,0.1573803,0.002160683,0.0007818451,0.0003971776,0.0002700443,0.00007991,0.002519354],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.936252,"threshold_uncertainty_score":0.9996998,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08598030702419207,"score_gpt":0.3185981659794001,"score_spread":0.232617858955208,"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."}}