{"id":"W7125949704","doi":"10.1109/smc58881.2025.11343334","title":"Phonetic Analysis of Real and Synthetic Speech Using HuBERT Embeddings: Perspectives for Deepfake Detection","year":2025,"lang":"en","type":"article","venue":"","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"Institut National de la Recherche Scientifique; Cégep de l'Outaouais","funders":"","keywords":"Speech synthesis; Synthetic data; Sophistication; Speech processing; Rank (graph theory); Divergence (linguistics); Noise (video)","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":[],"consensus_categories":[],"category_scores_codex":[0.0002367955,0.00009145964,0.0002351485,0.0006545513,0.0001058115,0.00007166711,0.0001547874,0.00005282607,0.00003691025],"category_scores_gemma":[0.0001253707,0.00008408112,0.0001356461,0.001225512,0.00005026949,0.000138425,0.00005686827,0.00003394444,9.376281e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00004750668,"about_ca_system_score_gemma":0.00002561441,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002024887,"about_ca_topic_score_gemma":0.0001989287,"domain_scores_codex":[0.9991947,0.0000406875,0.0001957044,0.0003256358,0.0001096947,0.0001335695],"domain_scores_gemma":[0.9992581,0.0002636999,0.00006768203,0.0002219416,0.0001537561,0.00003479521],"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":[0.00004911662,0.0001776496,0.001177675,0.00007594396,0.001245523,0.000002492041,0.002183186,0.0001162676,0.08438177,0.0155384,0.00002821378,0.8950238],"study_design_scores_gemma":[0.0002241482,0.00004571121,0.005465728,0.00002991604,0.0005781496,0.000006760073,0.001432419,0.8155866,0.1740449,0.002399127,0.00004734016,0.0001392369],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3562748,0.00006583663,0.6398426,0.00008642569,0.00006071242,0.0001210865,0.000001976814,0.00005414692,0.003492414],"genre_scores_gemma":[0.8766727,0.00004728112,0.1229964,0.0000373268,0.000007804215,0.000008529008,4.38588e-7,0.000003503895,0.0002260269],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8948845,"threshold_uncertainty_score":0.3428729,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01738696341447149,"score_gpt":0.2900737172972532,"score_spread":0.2726867538827817,"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."}}