{"id":"W3209371554","doi":"10.1109/icassp43922.2022.9747760","title":"Optimizing Alignment of Speech and Language Latent Spaces for End-To-End Speech Recognition and Understanding","year":2022,"lang":"en","type":"article","venue":"ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Computer science; Speech recognition; Encoder; Language model; Embedding; Word error rate; Context (archaeology); Connectionism; End-to-end principle; Task (project management); Natural language processing; Artificial intelligence; Artificial neural network","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"],"consensus_categories":[],"category_scores_codex":[0.001068662,0.0003490058,0.0004299476,0.0005630247,0.0005371586,0.0006623314,0.0005410781,0.00009936922,0.000805891],"category_scores_gemma":[0.000157221,0.0003626558,0.00008929951,0.0003238334,0.0001616086,0.000421311,0.0004216921,0.0004127426,0.000006770495],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002967579,"about_ca_system_score_gemma":0.0001672884,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006215956,"about_ca_topic_score_gemma":0.00001842226,"domain_scores_codex":[0.9968516,0.0001387062,0.000573766,0.0008841428,0.001120298,0.0004314599],"domain_scores_gemma":[0.9983953,0.0004385238,0.0003890551,0.0002189545,0.0002925251,0.0002656376],"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":[0.000401212,0.0003189135,0.000142217,0.0002188,0.0001900813,0.0002740487,0.002637953,0.0001815547,0.1624991,0.003255333,0.001035107,0.8288457],"study_design_scores_gemma":[0.004133193,0.00266385,0.0003103774,0.0009125327,0.0003008949,0.001266505,0.02343434,0.7990626,0.114392,0.04958993,0.001803517,0.002130356],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.40175,0.0006543219,0.5790682,0.008729313,0.001449397,0.00153112,0.001453741,0.000271251,0.005092709],"genre_scores_gemma":[0.9208187,0.000235321,0.07748774,0.0005894297,0.0001801859,0.00009622949,0.00007810635,0.00003368888,0.0004805829],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8267153,"threshold_uncertainty_score":0.9998825,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1064410328063218,"score_gpt":0.3075885950457152,"score_spread":0.2011475622393934,"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."}}