{"id":"W4388912435","doi":"10.3390/app132312571","title":"Environment-Aware Knowledge Distillation for Improved Resource-Constrained Edge Speech Recognition","year":2023,"lang":"en","type":"article","venue":"Applied Sciences","topic":"Speech and Audio Processing","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université du Québec en Outaouais; Institut National de la Recherche Scientifique; Université du Québec à Montréal","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Speech recognition; Word error rate; Estimator; Benchmark (surveying); Inference; Artificial intelligence; Noise reduction; Machine learning","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008164481,0.0001362469,0.0001304372,0.000179343,0.0006269261,0.000278457,0.0006077767,0.00006241825,0.00001581703],"category_scores_gemma":[0.00004746848,0.0001211392,0.00004973532,0.000900773,0.0002460376,0.0003661476,0.0001519009,0.00006775018,0.0002703432],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003302837,"about_ca_system_score_gemma":0.0000883702,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000001456583,"about_ca_topic_score_gemma":0.000002693185,"domain_scores_codex":[0.9985749,0.00001765549,0.0002065831,0.0005888368,0.0001935349,0.0004184586],"domain_scores_gemma":[0.9993619,0.0002127082,0.00010661,0.0002068155,0.00002675675,0.00008525497],"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":[0.000006055083,0.00002268018,0.00005222647,0.00001945865,0.000004478009,7.509018e-7,0.0003937944,0.00003212837,0.05014019,0.0006679406,0.001472198,0.9471881],"study_design_scores_gemma":[0.001487677,0.0002850307,0.001578569,0.00007178773,0.00002342451,0.00001797914,0.001074956,0.1399296,0.7191237,0.09743183,0.03796895,0.001006499],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1964595,0.0001334164,0.7579233,0.002595546,0.0006391898,0.001405177,0.00004447522,0.001378464,0.03942099],"genre_scores_gemma":[0.9092757,0.00001263612,0.08956614,0.0001902423,0.0002449948,0.0001292782,0.00004958035,0.00001276267,0.0005186741],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9461816,"threshold_uncertainty_score":0.4939913,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.039738359534765,"score_gpt":0.2665861711993528,"score_spread":0.2268478116645878,"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."}}