{"id":"W3207551191","doi":"10.1109/icassp43922.2022.9747180","title":"MetricGAN-U: Unsupervised Speech Enhancement/ Dereverberation Based Only on Noisy/ Reverberated Speech","year":2022,"lang":"en","type":"article","venue":"ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","topic":"Speech and Audio Processing","field":"Computer Science","cited_by":40,"is_retracted":false,"has_abstract":true,"ca_institutions":"Mila - Quebec Artificial Intelligence Institute","funders":"","keywords":"Computer science; Constraint (computer-aided design); Speech enhancement; Speech recognition; Noise (video); Artificial intelligence; Deep learning; Noise measurement; Unsupervised learning; Noise reduction; Image (mathematics); Mathematics","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","sts","scholarly_communication","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001525006,0.0008292431,0.0007233726,0.001189008,0.001574004,0.001920374,0.002323915,0.0002296612,0.003566406],"category_scores_gemma":[0.0003435008,0.0008641952,0.0002108523,0.001902098,0.0001926566,0.00107214,0.0005555788,0.001706363,0.0001583262],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008501781,"about_ca_system_score_gemma":0.001473373,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006523768,"about_ca_topic_score_gemma":0.00001515493,"domain_scores_codex":[0.9921465,0.0004048875,0.001161046,0.001928883,0.003347188,0.001011512],"domain_scores_gemma":[0.9965362,0.0004192431,0.0007864962,0.0007397152,0.001067042,0.0004512955],"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":[0.000695296,0.001630326,0.0003491549,0.0001820068,0.0002011555,0.00117223,0.0004068864,0.002990542,0.3987908,0.00162547,0.01255449,0.5794017],"study_design_scores_gemma":[0.003638997,0.002131582,0.0003280591,0.0004545961,0.0001254148,0.0004584834,0.0005853157,0.6792112,0.2978686,0.006252001,0.006951248,0.001994479],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2209495,0.0009032606,0.7284918,0.01537054,0.004870527,0.001865853,0.0005336138,0.001119619,0.02589525],"genre_scores_gemma":[0.9373222,0.0001540548,0.05111038,0.005492317,0.0005607963,0.0001343725,0.0002824262,0.0000887722,0.004854705],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7163727,"threshold_uncertainty_score":0.9997258,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04401044470129294,"score_gpt":0.2939940872545188,"score_spread":0.2499836425532259,"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."}}