{"id":"W4408354411","doi":"10.1109/icassp49660.2025.10888241","title":"LAVCap: LLM-based Audio-Visual Captioning using Optimal Transport","year":2025,"lang":"en","type":"article","venue":"","topic":"Music and Audio Processing","field":"Computer Science","cited_by":6,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Closed captioning; Computer science; Audio visual; Speech recognition; Multimedia; Artificial intelligence; Image (mathematics)","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001352463,0.003396415,0.001311176,0.00169672,0.0008369017,0.001995081,0.003338553,0.002341054,0.01476768],"category_scores_gemma":[0.004755554,0.0008199926,0.002035897,0.001336976,0.001020031,0.003151606,0.00444997,0.003335099,0.01042137],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001063409,"about_ca_system_score_gemma":0.001281066,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004223759,"about_ca_topic_score_gemma":0.006954331,"domain_scores_codex":[0.9988037,0.0003012061,0.00005945943,0.0003961935,0.000308529,0.0001309529],"domain_scores_gemma":[0.9987306,0.0004520985,0.00008226449,0.0003087684,0.0003114662,0.0001148392],"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":[0.0007324429,0.0003225029,0.0005774039,0.001089298,0.0001777777,0.000535763,0.0003554081,0.06978088,0.06098185,0.006374268,0.1040422,0.7550303],"study_design_scores_gemma":[0.0001087557,0.000256665,0.0005128856,0.0000913554,0.00008045242,0.0004364378,0.0001678747,0.9027887,0.04686193,0.01184793,0.03674414,0.0001028422],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.009286377,0.001693401,0.9152028,0.0004510678,0.0007863261,0.0004315624,0.002578898,0.06279496,0.006774619],"genre_scores_gemma":[0.1783395,0.001380023,0.7709436,0.001445259,0.0006397386,0.001099427,0.01936225,0.00523802,0.02155211],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01476768,"threshold_uncertainty_score":0.04940277,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01577899618749308,"score_gpt":0.2860186404324295,"score_spread":0.2702396442449365,"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."}}