{"id":"W4406080107","doi":"10.1016/j.rineng.2025.103943","title":"Automated speech therapy through personalized pronunciation correction using reinforcement learning and large language models","year":2025,"lang":"en","type":"article","venue":"Results in Engineering","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Pronunciation; Computer science; Reinforcement learning; Natural language processing; Reinforcement; Speech therapy; Speech recognition; Artificial intelligence; Linguistics; Psychology; Audiology; Medicine; Social psychology","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000995112,0.001063194,0.0007460542,0.0003946373,0.0003314796,0.00105302,0.001287409,0.0007407431,0.003997616],"category_scores_gemma":[0.003425376,0.0003424117,0.0006286255,0.0002547877,0.0004354556,0.0009469719,0.001436319,0.001292654,0.002698501],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005261144,"about_ca_system_score_gemma":0.001107275,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003175842,"about_ca_topic_score_gemma":0.003212302,"domain_scores_codex":[0.9990358,0.0002672604,0.00006324764,0.0003302858,0.000247489,0.0000559566],"domain_scores_gemma":[0.9992173,0.000386552,0.00007230063,0.0001355218,0.0001353494,0.00005299226],"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":[0.0004653204,0.00036465,0.002719106,0.0002673137,0.0001330351,0.0003809431,0.0004430142,0.1170601,0.05434746,0.002611995,0.006342474,0.8148646],"study_design_scores_gemma":[0.00006420612,0.0001921772,0.0009494812,0.00003055912,0.00004413828,0.0003020991,0.00008652841,0.9615126,0.0255884,0.004144338,0.007036712,0.00004878311],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02375467,0.0003241958,0.9557244,0.000301335,0.0001169527,0.000203867,0.0002340717,0.01672679,0.002613581],"genre_scores_gemma":[0.5421859,0.0002853199,0.4472641,0.0003424064,0.00006848174,0.0005093085,0.0007219429,0.0006885157,0.007933928],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.003997616,"threshold_uncertainty_score":0.01337337,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01881285815148709,"score_gpt":0.2752923702156232,"score_spread":0.2564795120641362,"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."}}