{"id":"W3083173864","doi":"10.1145/3414685.3417838","title":"Speech gesture generation from the trimodal context of text, audio, and speaker identity","year":2020,"lang":"en","type":"article","venue":"ACM Transactions on Graphics","topic":"Human Pose and Action Recognition","field":"Computer Science","cited_by":304,"is_retracted":false,"has_abstract":true,"ca_institutions":"Kootenay Association for Science & Technology","funders":"","keywords":"Gesture; Computer science; Context (archaeology); Speech recognition; Identity (music); Gesture recognition; Embedding; Artificial intelligence; Human–computer interaction; Natural language processing","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":[0.0004349536,0.000810434,0.0004113978,0.0001934275,0.0002186092,0.0003302261,0.0005854629,0.0006159085,0.002343138],"category_scores_gemma":[0.001681223,0.0002400102,0.0004976796,0.0001275841,0.0004992837,0.0005280126,0.001027767,0.0009310787,0.0007376215],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003021893,"about_ca_system_score_gemma":0.0004009242,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001659468,"about_ca_topic_score_gemma":0.003444768,"domain_scores_codex":[0.999694,0.00007342759,0.00001228582,0.0001153327,0.000074318,0.0000306432],"domain_scores_gemma":[0.9996254,0.0001941105,0.00002923593,0.00008383472,0.00003808774,0.00002931623],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000499445,0.0001629853,0.002068994,0.0002037431,0.0000621363,0.0005861355,0.0002739342,0.5579667,0.1021224,0.007562345,0.003253141,0.3252379],"study_design_scores_gemma":[0.00001672766,0.000144862,0.001072443,0.00002196755,0.00001714222,0.000241408,0.00003030727,0.9703842,0.02248691,0.003566318,0.00199892,0.00001876647],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.07364281,0.0005106358,0.9195485,0.0002087288,0.0001348963,0.0001406486,0.0001986785,0.00184934,0.003765821],"genre_scores_gemma":[0.8638378,0.0003470803,0.1283135,0.0002228367,0.0000616211,0.0001949505,0.0004888679,0.0001599126,0.006373513],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.002343138,"threshold_uncertainty_score":0.007838607,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0463744464987958,"score_gpt":0.253126340540515,"score_spread":0.2067518940417192,"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."}}