{"id":"W4319862473","doi":"10.1109/slt54892.2023.10022470","title":"A Comprehensive Study on Self-Supervised Distillation for Speaker Representation Learning","year":2023,"lang":"en","type":"article","venue":"2022 IEEE Spoken Language Technology Workshop (SLT)","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Computer science; Benchmark (surveying); Representation (politics); Speaker recognition; Speech recognition; Artificial intelligence; Word error rate; Training set; Feature learning; Machine learning","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003605059,0.001065298,0.001017383,0.0006946272,0.0004862181,0.0011023,0.001354877,0.001165771,0.00177282],"category_scores_gemma":[0.008264349,0.0005193373,0.001037522,0.0007800667,0.001135928,0.002674512,0.001735883,0.002831558,0.0008161052],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007435746,"about_ca_system_score_gemma":0.001043297,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001944961,"about_ca_topic_score_gemma":0.001764342,"domain_scores_codex":[0.9976721,0.001023852,0.00008962523,0.0006526966,0.0004602716,0.0001013065],"domain_scores_gemma":[0.9956799,0.002698632,0.0001316641,0.0007178902,0.0006687007,0.000103273],"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":[0.0003732503,0.0004380645,0.001499041,0.0006666543,0.0003113221,0.0001082448,0.0003100211,0.2219949,0.02415021,0.03697198,0.007309543,0.7058668],"study_design_scores_gemma":[0.00001001232,0.0001919766,0.0005754688,0.00004438415,0.00003622156,0.00009654514,0.00003062885,0.9750413,0.009256453,0.009560162,0.005128109,0.00002867056],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02308675,0.009466736,0.9612085,0.000652349,0.0001621032,0.00009059976,0.00009382643,0.0009831393,0.004255979],"genre_scores_gemma":[0.5702087,0.009813893,0.4021079,0.001127949,0.0008895111,0.0003125748,0.001398523,0.0006469625,0.01349394],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003605059,"threshold_uncertainty_score":0.01906556,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03729401784374337,"score_gpt":0.322562182637299,"score_spread":0.2852681647935556,"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."}}