{"id":"W4415524446","doi":"10.1109/mlsp62443.2025.11204268","title":"Tiny Noise-Robust Voice Activity Detector for Voice Assistants","year":2025,"lang":"","type":"article","venue":"","topic":"Speech and Audio Processing","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Voice activity detection; Background noise; Noise (video); Detector; Human voice; Speech enhancement; SIGNAL (programming language)","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","scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.0008582218,0.0005738089,0.0006259115,0.0003185671,0.001099774,0.001515379,0.001884517,0.0003295716,0.0001057545],"category_scores_gemma":[0.0006977722,0.0005572679,0.0003352441,0.001904366,0.0001875511,0.001556414,0.0007299022,0.0004533298,0.0002362388],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002793942,"about_ca_system_score_gemma":0.001261067,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00008663417,"about_ca_topic_score_gemma":0.0002755365,"domain_scores_codex":[0.9958928,0.0001275235,0.0006114877,0.001554213,0.0005338927,0.001280137],"domain_scores_gemma":[0.996788,0.0008182696,0.0003621309,0.001139387,0.0005764368,0.0003157249],"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":[0.0002498364,0.0004943191,0.001165624,0.0008126454,0.0001706878,0.00001485789,0.0002425003,0.0001229986,0.08680236,0.0004982087,0.009266167,0.9001598],"study_design_scores_gemma":[0.001966493,0.000248762,0.01395267,0.000533717,0.0001358267,0.000009409637,0.00007514662,0.1195163,0.835045,0.001436932,0.02615203,0.0009277081],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04980921,0.0007466832,0.921759,0.004425893,0.003155841,0.0007506984,0.00002906765,0.0003398648,0.01898373],"genre_scores_gemma":[0.8350133,0.00003490192,0.1354243,0.002028974,0.0003543791,0.00007258068,0.000001454117,0.00003243039,0.02703768],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8992321,"threshold_uncertainty_score":0.9996879,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03097398684967214,"score_gpt":0.2877127385676779,"score_spread":0.2567387517180058,"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."}}