{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001093209,0.001543687,0.000823985,0.000577522,0.0003929079,0.0009043791,0.001008252,0.001278706,0.004214373],"category_scores_gemma":[0.003103561,0.0004707378,0.0008436767,0.0006672534,0.000597094,0.002260559,0.001730274,0.002242699,0.003730703],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005682614,"about_ca_system_score_gemma":0.00126468,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003741579,"about_ca_topic_score_gemma":0.0061799,"domain_scores_codex":[0.9989378,0.0003229029,0.00005724841,0.0003747347,0.0001844531,0.0001228995],"domain_scores_gemma":[0.99919,0.0003736384,0.000068635,0.0001424135,0.0001693658,0.00005595046],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.0005096679,0.0004596805,0.0012124,0.0001469032,0.0001045295,0.0002217159,0.0002761795,0.1307867,0.09805091,0.006518027,0.007201493,0.7545118],"study_design_scores_gemma":[0.0000186075,0.00009309261,0.0005372826,0.000009189198,0.00002527029,0.00008132283,0.00008022859,0.9654831,0.02680357,0.005384998,0.001465631,0.0000176011],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01812629,0.0002560626,0.9766309,0.0001549305,0.00005806689,0.00004953139,0.0001431654,0.003658449,0.0009226035],"genre_scores_gemma":[0.4726577,0.0004756964,0.5143723,0.0004207501,0.0001309392,0.0003437103,0.002077847,0.0008214117,0.008699557],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004214373,"threshold_uncertainty_score":0.01409847,"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."}}