{"id":"W4391831786","doi":"10.4995/eurocall2023.2023.17007","title":"Evaluating the effectiveness of Microsoft Transcribe for automating the assessment of pronunciation in language proficiency tests","year":2023,"lang":"en","type":"article","venue":"","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Concordia University","funders":"Université du Québec à Montréal","keywords":"Computer science; Pronunciation; Natural language processing; Test (biology); Automation; Reliability (semiconductor); Artificial intelligence; Linguistics; Engineering","routes":{"ca_aff":true,"ca_fund":true,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005536079,0.00005583926,0.000109775,0.00006972848,0.00007109073,0.00002523725,0.0003392533,0.00002127703,0.000004688017],"category_scores_gemma":[0.0006650136,0.00002995767,0.00005527441,0.0006186112,0.00002541218,0.00009668172,0.00003603933,0.00004468803,0.000001312968],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002708978,"about_ca_system_score_gemma":0.00009755242,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003687872,"about_ca_topic_score_gemma":0.00002354077,"domain_scores_codex":[0.9988719,0.0003731911,0.0002367749,0.000140394,0.0002533933,0.0001243363],"domain_scores_gemma":[0.9966709,0.002909775,0.0001091146,0.0001952146,0.0001069292,0.000008050678],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00000975753,0.000103784,0.007749006,0.0003563767,0.00001386422,4.643322e-7,0.004023521,0.0006750716,0.7129039,0.009993286,0.00001050059,0.2641605],"study_design_scores_gemma":[0.0002617065,0.0001059454,0.2955482,0.00009863186,0.000005295388,8.136336e-7,0.0005023252,0.5501445,0.1522676,0.001020057,0.000001100732,0.00004385077],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8749464,0.00001242784,0.1225773,0.0003429447,0.00005949053,0.001100991,0.000003653683,0.00006773073,0.0008890629],"genre_scores_gemma":[0.9651853,6.29434e-7,0.0345239,0.00002435358,0.000006000045,0.0002269265,0.000001761676,0.000003709857,0.00002735321],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.5606363,"threshold_uncertainty_score":0.1918706,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0713846181357021,"score_gpt":0.4181764047821827,"score_spread":0.3467917866464806,"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."}}