{"id":"W4416143831","doi":"10.65148/ecn/2025019","title":"Personalized Text to Speech Synthesis through Few Shot Speaker Adaptation with Contrastive Learning","year":2025,"lang":"en","type":"article","venue":"Elaris Computing Nexus","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"Trinity College","funders":"","keywords":"Naturalness; Similarity (geometry); Speaker recognition; Mean opinion score; Speech synthesis; Encoder; Feature learning; Word error rate; Speaker diarisation","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"],"consensus_categories":[],"category_scores_codex":[0.0005286428,0.0002900348,0.0004080866,0.0002336105,0.0004768646,0.0004135566,0.0005804159,0.0001073876,0.000116132],"category_scores_gemma":[0.0006886112,0.0002623815,0.0001149771,0.001111245,0.00007367441,0.0003230009,0.0002084934,0.0003552992,0.0002527627],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001264286,"about_ca_system_score_gemma":0.0001640309,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001739426,"about_ca_topic_score_gemma":0.00003857401,"domain_scores_codex":[0.9976974,0.0002959959,0.000353052,0.000729697,0.0004432402,0.0004806287],"domain_scores_gemma":[0.9978153,0.001282925,0.0001558518,0.0003571288,0.0002664168,0.0001223675],"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":[0.0001201471,0.0001566053,0.0008683947,0.00003633078,0.000247112,0.0001250375,0.007817047,0.003234201,0.0005639483,0.02382348,0.00169816,0.9613096],"study_design_scores_gemma":[0.00316012,0.0005983626,0.01532948,0.002062491,0.0002778177,0.0002607374,0.008620107,0.859576,0.03244366,0.004119351,0.07147193,0.002079884],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02263381,0.0001162004,0.9205774,0.002026023,0.0003398126,0.000352365,0.00000224597,0.0004641877,0.05348794],"genre_scores_gemma":[0.7450486,0.000005038651,0.2520107,0.001451155,0.00008119529,0.00001764244,0.000003088528,0.00001938585,0.001363113],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9592296,"threshold_uncertainty_score":0.9999828,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02728395361844038,"score_gpt":0.2702029549770197,"score_spread":0.2429190013585793,"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."}}