{"id":"W7148418289","doi":"10.1109/asru65441.2025.11434627","title":"Low-Resource Domain Adaptation for Speech LLMs via Text-Only Fine-Tuning","year":2025,"lang":"","type":"article","venue":"","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"Optech (Canada)","funders":"","keywords":"Adaptation (eye); Domain (mathematical analysis); Generalization; Domain adaptation; Encoder; Mechanism (biology); Degradation (telecommunications)","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","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001186849,0.0004667969,0.0005149624,0.0006010969,0.0006652804,0.0007477919,0.001151126,0.0003018936,0.000998765],"category_scores_gemma":[0.0004020112,0.000471801,0.0004121588,0.001396604,0.0001341859,0.0005865784,0.0003206559,0.0002720952,0.0004400703],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001582971,"about_ca_system_score_gemma":0.0004801065,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006684341,"about_ca_topic_score_gemma":0.0002814326,"domain_scores_codex":[0.9963968,0.0002484,0.0009026531,0.001118902,0.0005299672,0.0008032151],"domain_scores_gemma":[0.9970114,0.001253773,0.0002636382,0.0008437685,0.0003916482,0.0002357939],"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":[0.00007109751,0.0002228007,0.00005748915,0.0001110302,0.0000876017,0.00001528092,0.0006480032,0.00006737757,0.00240126,0.02137414,0.005351286,0.9695926],"study_design_scores_gemma":[0.001941298,0.0002319186,0.0004919983,0.0005104976,0.00009213994,0.00005165071,0.001166136,0.8318667,0.05017357,0.0226208,0.09004719,0.0008061393],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.00601823,0.0001714706,0.916426,0.005966059,0.001043303,0.0009793171,0.00001644695,0.0002667022,0.06911248],"genre_scores_gemma":[0.2452197,0.00003185918,0.6949939,0.005204828,0.0003690841,0.0001253063,0.0000320354,0.00004979673,0.05397349],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.9687865,"threshold_uncertainty_score":0.9999145,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02317349662740988,"score_gpt":0.2601340969287989,"score_spread":0.236960600301389,"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."}}