{"meta":{"query_hash":"a8a6aac901d9","filters":{"venue":"IEEE Transactions on Speech and Audio Processing"},"cohort_total":16,"direct_labels_cover":0,"predictions_cover":16,"exported":16,"export_cap":100000,"truncated":false,"label_status":"direct model label, unvalidated","prediction_status":"machine_predicted_unvalidated (Codex and Gemma teacher distillation)","score_status":"score_only:v0-immature-baseline","snapshot":{"source":"OpenAlex, pinned release, all 482 partitions","release":"2026-06-24","frame_built":"2026-07-12"},"permalink":"https://metacan.xera.ac/q/a8a6aac901d9","api":"https://metacan.xera.ac/api/v1/cohort?venue=IEEE+Transactions+on+Speech+and+Audio+Processing"},"results":[{"id":"W2000916836","doi":"10.1109/89.902276","title":"An adaptive KLT approach for speech enhancement","year":2001,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Speech and Audio Processing","field":"Computer Science","cited_by":235,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Toronto","funders":"","keywords":"Speech enhancement; Speech recognition; Noise (video); Computer science; Speech processing; White noise; Additive white Gaussian noise; Residual; Background noise; Colors of noise; Distortion (music); Mathematics; Artificial intelligence; Noise reduction; Algorithm; Telecommunications","score_opus":0.027864317274008104,"score_gpt":0.2735090793971463,"score_spread":0.2456447621231382,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2000916836","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.001977151,0.00019757835,0.99688905,0.000032648513,0.000025128304,0.000015225395,0.00000878919,0.00025767725,0.0005967661],"genre_scores_gemma":[0.08430077,0.00063711114,0.9094498,0.00012436822,0.00008878113,0.00007807832,0.00012264411,0.00012493966,0.0050734826],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995084,0.00007657913,0.000029937042,0.00010271587,0.00024999824,0.000032392945],"domain_scores_gemma":[0.99969554,0.00009743718,0.000027432878,0.000035281966,0.00013200333,0.000012287406],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00046291828,0.0007296977,0.00051983155,0.00074408785,0.00026572894,0.0005898561,0.00068969635,0.0006557532,0.0020789753],"category_scores_gemma":[0.0009670551,0.00024831863,0.00068859133,0.0005719202,0.00043792767,0.0008595167,0.0006135628,0.00075173966,0.0013086026],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019908903,0.00007506512,0.00040371035,0.00016606125,0.00006520696,0.00015664086,0.00013208028,0.05082987,0.20244932,0.010432391,0.0016153544,0.7334752],"study_design_scores_gemma":[0.000028400182,0.00021584517,0.0008845259,0.000019916482,0.000056338537,0.0006063598,0.000043272365,0.8977051,0.080752015,0.0058645695,0.013775925,0.000047764308],"about_ca_topic_score_codex":0.00087115524,"about_ca_topic_score_gemma":0.0013683243,"teacher_disagreement_score":0.0020789753,"about_ca_system_score_codex":0.00030307221,"about_ca_system_score_gemma":0.00041036785,"threshold_uncertainty_score":0.0069549084},"labels":[],"label_agreement":null},{"id":"W2096441936","doi":"10.1109/tsa.2005.851925","title":"A frequency domain method for blind source separation of convolutive audio mixtures","year":2005,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Blind Source Separation Techniques","field":"Computer Science","cited_by":132,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McMaster University","funders":"","keywords":"Blind signal separation; Algorithm; Initialization; Frequency domain; Computer science; Mixing (physics); Joint (building); Bin; Speech recognition; Permutation (music); Spectral density; Mathematics; Acoustics","score_opus":0.019362324051516985,"score_gpt":0.32183531275924726,"score_spread":0.30247298870773026,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2096441936","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0005050144,0.00014763983,0.9988146,0.000023897863,0.000048189173,0.000011308506,0.000010394423,0.00012496389,0.00031411965],"genre_scores_gemma":[0.013794181,0.0003490565,0.98343223,0.0000529528,0.00009109887,0.00006768737,0.00007645104,0.000045665063,0.0020906492],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9994382,0.000110739806,0.0000274311,0.000110966044,0.00029230278,0.000020347714],"domain_scores_gemma":[0.9996153,0.00012730775,0.000031524345,0.000057073907,0.00015054233,0.00001823256],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00057739514,0.00086181494,0.0006171268,0.0009111896,0.0004407159,0.0005406355,0.00071613374,0.00096906803,0.0027231341],"category_scores_gemma":[0.0011959517,0.0003341518,0.0006145428,0.000648643,0.00064539054,0.0010296117,0.0006566234,0.0012141353,0.0023027698],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00019571777,0.00009942663,0.0002579902,0.00034585156,0.00010306326,0.00011523554,0.00008983287,0.057314698,0.13417625,0.034487568,0.003721916,0.76909256],"study_design_scores_gemma":[0.000073374016,0.00021353045,0.00064576796,0.000037474383,0.00005972961,0.0010361788,0.000035435503,0.8668178,0.06585313,0.015253846,0.04986159,0.000112133246],"about_ca_topic_score_codex":0.00077058683,"about_ca_topic_score_gemma":0.0010762449,"teacher_disagreement_score":0.0027231341,"about_ca_system_score_codex":0.00030221354,"about_ca_system_score_gemma":0.0005980161,"threshold_uncertainty_score":0.009109795},"labels":[],"label_agreement":null},{"id":"W2100555417","doi":"10.1109/tsa.2003.815518","title":"A soft voice activity detector based on a laplacian-gaussian model","year":2003,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Speech and Audio Processing","field":"Computer Science","cited_by":141,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"","keywords":"Speech recognition; Hidden Markov model; Computer science; Noise (video); Discrete cosine transform; Gaussian; Posterior probability; Detector; Probability distribution; Bayesian probability; Pattern recognition (psychology); Mathematics; Artificial intelligence; Statistics","score_opus":0.01821155687566045,"score_gpt":0.24881046777282215,"score_spread":0.2305989108971617,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2100555417","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006184169,0.00021325046,0.99237305,0.000065138156,0.00004330728,0.000022041197,0.000036435915,0.00044669025,0.00061597244],"genre_scores_gemma":[0.5598485,0.0006526265,0.43310013,0.00043324052,0.0001583067,0.00014608912,0.00036317084,0.00010658814,0.005191373],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9992642,0.00016534339,0.000032413333,0.00017848735,0.00029254577,0.000066991124],"domain_scores_gemma":[0.99904424,0.0005858412,0.000058441743,0.00006868667,0.00019465706,0.00004809914],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009163311,0.0005790011,0.0010279054,0.00094351714,0.00026436965,0.0010049671,0.0011508415,0.0010061407,0.0015050998],"category_scores_gemma":[0.0023669514,0.00036039914,0.00067360094,0.00055837113,0.00057544914,0.0013557728,0.00083384,0.0008663945,0.0008030797],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00057373015,0.00018820915,0.004332872,0.00024968793,0.00020728492,0.0003746977,0.00016545875,0.19579095,0.07481424,0.024198374,0.0037571588,0.69534737],"study_design_scores_gemma":[0.000013127591,0.000055266268,0.00058382837,0.000005697992,0.000017335395,0.000168727,0.000009189828,0.98985404,0.005208374,0.0030422348,0.0010189754,0.000023223618],"about_ca_topic_score_codex":0.0018408606,"about_ca_topic_score_gemma":0.0019802274,"teacher_disagreement_score":0.0018408606,"about_ca_system_score_codex":0.0005112033,"about_ca_system_score_gemma":0.0005900709,"threshold_uncertainty_score":0.0050351024},"labels":[],"label_agreement":null},{"id":"W2101999036","doi":"10.1109/tsa.2005.851943","title":"Speech enhancement employing Laplacian-Gaussian mixture","year":2005,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Speech and Audio Processing","field":"Computer Science","cited_by":68,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"","keywords":"Speech recognition; Wiener filter; Computer science; Estimator; Minimum mean square error; Speech enhancement; Gaussian; Computational complexity theory; Filter (signal processing); Noise (video); Algorithm; Mathematics; Artificial intelligence; Statistics; Physics","score_opus":0.013688057152124792,"score_gpt":0.25468810587074936,"score_spread":0.24100004871862457,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2101999036","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006099049,0.00022596988,0.99255717,0.000038865343,0.000020726937,0.000008893188,0.0000077947825,0.0003530012,0.0006884689],"genre_scores_gemma":[0.19274545,0.0006515088,0.80176604,0.00013138361,0.00006087658,0.000032495325,0.0001209022,0.00009163809,0.004399719],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9997763,0.000043667278,0.000010623603,0.000037238464,0.00011624336,0.00001604136],"domain_scores_gemma":[0.9997845,0.00009775748,0.000017506454,0.000025210471,0.00006545375,0.000009584506],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00044863837,0.00047214096,0.00048611243,0.0005324307,0.00018174811,0.0003897675,0.0004357215,0.00051031116,0.00093854964],"category_scores_gemma":[0.0008865066,0.0002326755,0.00057572866,0.00043239546,0.00028122988,0.00082462083,0.00066491385,0.00042599972,0.0006423585],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00028109035,0.00008452674,0.001127435,0.000158582,0.00012528418,0.00020305166,0.00014108114,0.15681152,0.1599396,0.014470479,0.0025737504,0.6640836],"study_design_scores_gemma":[0.000017092661,0.000066445304,0.00050331064,0.0000068190566,0.000032426495,0.00022242354,0.000012500359,0.96346486,0.02854105,0.0024284988,0.0046804943,0.000024028386],"about_ca_topic_score_codex":0.001030732,"about_ca_topic_score_gemma":0.0018179977,"teacher_disagreement_score":0.001030732,"about_ca_system_score_codex":0.00021931593,"about_ca_system_score_gemma":0.00029704403,"threshold_uncertainty_score":0.0031397939},"labels":[],"label_agreement":null},{"id":"W2121415728","doi":"10.1109/tsa.2004.840940","title":"Eigenvoice modeling with sparse training data","year":2005,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":472,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Computer Research Institute of Montréal","funders":"","keywords":"Computer science; Speech recognition; Training set; Set (abstract data type); Adaptation (eye); Limit (mathematics); Training (meteorology); Maximum likelihood; Covariance matrix; Pattern recognition (psychology); Estimation theory; Artificial intelligence; Covariance; Speaker recognition; Algorithm; Mathematics; Statistics","score_opus":0.095794255220844,"score_gpt":0.2797844726229301,"score_spread":0.1839902174020861,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2121415728","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005680046,0.00006269001,0.99351764,0.000043627537,0.000013688858,0.000009312174,0.000045552093,0.00014377378,0.00048372726],"genre_scores_gemma":[0.47542048,0.00054984255,0.51559895,0.00018576637,0.0001511257,0.00022798829,0.00079956354,0.0002749555,0.006791252],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995765,0.000118992095,0.000016325788,0.000101235,0.00012877234,0.000058186564],"domain_scores_gemma":[0.99912506,0.00044245063,0.00008881113,0.00015130657,0.00016274443,0.000029627281],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007939901,0.00053049513,0.0007143595,0.00049459544,0.00026938258,0.0007678655,0.0013526723,0.00080754247,0.0014852852],"category_scores_gemma":[0.003208085,0.0005329204,0.00067358714,0.00056437205,0.00056212745,0.0012259965,0.0010610936,0.0012674484,0.00069066195],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001051985,0.000052041392,0.000860624,0.00009482021,0.0000609196,0.00010855857,0.00023275295,0.80146766,0.012447549,0.033431843,0.0022378552,0.14890021],"study_design_scores_gemma":[0.0000026325647,0.000007831561,0.00010531008,0.0000035166584,0.0000029667224,0.00002115068,0.000006456563,0.99129045,0.0010478433,0.006963778,0.0005419836,0.000006088376],"about_ca_topic_score_codex":0.00244556,"about_ca_topic_score_gemma":0.0042363885,"teacher_disagreement_score":0.00244556,"about_ca_system_score_codex":0.000251907,"about_ca_system_score_gemma":0.0004872123,"threshold_uncertainty_score":0.0049687624},"labels":[],"label_agreement":null},{"id":"W2127390035","doi":"10.1109/89.902283","title":"Synthetic stereo acoustic echo cancellation structure for multiple participant VoIP conferences","year":2001,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Speech and Audio Processing","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Carleton University","funders":"","keywords":"Monaural; Spatialization; Echo (communications protocol); Computer science; Stereophonic sound; Teleconference; Loudspeaker; Voice over IP; Reverberation; Speech recognition; Channel (broadcasting); Artificial intelligence; Acoustics; Telecommunications; The Internet","score_opus":0.0403782802899136,"score_gpt":0.2751404288449312,"score_spread":0.23476214855501756,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2127390035","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.08687777,0.00028469213,0.90526205,0.00016013977,0.0001044501,0.000061275634,0.00007407504,0.00071880774,0.0064567635],"genre_scores_gemma":[0.6702645,0.0001895027,0.32298318,0.00017502633,0.00011954613,0.000089099616,0.00026201148,0.00007317722,0.0058439868],"study_design_codex":"bench_or_experimental","study_design_gemma":"design_other","domain_scores_codex":[0.9996222,0.000086365326,0.000012802313,0.000035109115,0.00021407651,0.000029503097],"domain_scores_gemma":[0.9995733,0.000088288856,0.000037899954,0.00006188003,0.00018940997,0.000049133774],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00033824667,0.0002574263,0.00020430032,0.00025786992,0.00027024612,0.00030204645,0.00065176154,0.00046335292,0.0021040577],"category_scores_gemma":[0.0007470513,0.00012851966,0.00025846978,0.00019884763,0.00024898147,0.0004716597,0.00038222887,0.00035644046,0.0008215659],"study_design_candidate":"design_other","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0007187905,0.0001802876,0.0007267119,0.00015310226,0.00005011259,0.00027763163,0.00017538351,0.05866694,0.6360421,0.015967026,0.0023370655,0.28470474],"study_design_scores_gemma":[0.00015877324,0.001299353,0.0017962136,0.00002204078,0.00005926634,0.0011422159,0.00011908521,0.71428216,0.2516087,0.0052288948,0.024197701,0.000085682615],"about_ca_topic_score_codex":0.00040249096,"about_ca_topic_score_gemma":0.00076509465,"teacher_disagreement_score":0.0021040577,"about_ca_system_score_codex":0.00023869255,"about_ca_system_score_gemma":0.00042526348,"threshold_uncertainty_score":0.0070387125},"labels":[],"label_agreement":null},{"id":"W2128604088","doi":"10.1109/tsa.2003.818031","title":"Incorporating the human hearing properties in the signal subspace approach for speech enhancement","year":2003,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Speech and Audio Processing","field":"Computer Science","cited_by":171,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Speech recognition; Speech enhancement; Computer science; Signal subspace; Noise (video); Spectrogram; Noise reduction; Residual; Colors of noise; Subspace topology; Filter (signal processing); Artificial intelligence; Algorithm; Computer vision","score_opus":0.04549423966659918,"score_gpt":0.264615521974624,"score_spread":0.21912128230802483,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2128604088","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.005253729,0.00016250124,0.99368,0.000038099613,0.000019979023,0.000014303878,0.000009518229,0.0002532347,0.0005686007],"genre_scores_gemma":[0.14439824,0.0010962941,0.8510427,0.000073806,0.00006021561,0.00005615828,0.00009345928,0.000091080896,0.0030879872],"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99985766,0.000053465712,0.000008450925,0.000018543224,0.00005344889,0.000008488321],"domain_scores_gemma":[0.9998042,0.0001016985,0.00001157122,0.00003311524,0.00004353614,0.0000059890776],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00036318754,0.0005648311,0.0003603335,0.00030278572,0.00018719959,0.0004266967,0.00024412037,0.0005556873,0.0019957267],"category_scores_gemma":[0.0006798014,0.00020138458,0.00054053275,0.00027240434,0.0003328492,0.00083422574,0.00032760086,0.00041088538,0.0010251733],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001994317,0.0000951621,0.00036807623,0.000258198,0.000060166527,0.00028027903,0.00015861733,0.085805334,0.4519422,0.017887456,0.00086761796,0.44207755],"study_design_scores_gemma":[0.000026142692,0.0006423569,0.0010133641,0.00003118947,0.00006381241,0.0014875593,0.00007736367,0.7922491,0.17757984,0.012171857,0.014582166,0.00007534271],"about_ca_topic_score_codex":0.00039724546,"about_ca_topic_score_gemma":0.00083883625,"teacher_disagreement_score":0.0019957267,"about_ca_system_score_codex":0.000084638144,"about_ca_system_score_gemma":0.00026528543,"threshold_uncertainty_score":0.0066763163},"labels":[],"label_agreement":null},{"id":"W2132572274","doi":"10.1109/tsa.2005.851917","title":"LSP quantization by a union of locally trained codebooks","year":2005,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Advanced Data Compression Techniques","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"McGill University","funders":"","keywords":"Codebook; Vector quantization; Algorithm; Mathematics; Encoder; Speech coding; Computational complexity theory; Code-excited linear prediction; Speech recognition; Pattern recognition (psychology); Linear predictive coding; Computer science; Artificial intelligence; Statistics","score_opus":0.012466712998972692,"score_gpt":0.2608341671625037,"score_spread":0.24836745416353098,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2132572274","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0038536994,0.00009479481,0.99495727,0.00003997937,0.000016085964,0.000018721363,0.000033096025,0.00040185486,0.00058451836],"genre_scores_gemma":[0.2187703,0.00020732409,0.775421,0.00014736291,0.000060563325,0.00020078504,0.00034933264,0.00019141949,0.004651871],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.998968,0.00025914662,0.000064889486,0.00019867644,0.00045433117,0.000054958087],"domain_scores_gemma":[0.99878937,0.0003750206,0.00008447191,0.00038647288,0.0003199825,0.0000447737],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00118314,0.00053817924,0.0007995916,0.00041181003,0.0003084451,0.0007502843,0.0015131847,0.0006883538,0.0033875706],"category_scores_gemma":[0.0040501314,0.00038956173,0.00042596497,0.0007312929,0.0006800791,0.0019123844,0.0012670917,0.0011868813,0.0018575151],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00037302007,0.00009371246,0.0005050057,0.00013124642,0.000053649535,0.00007352538,0.00014865062,0.4807514,0.036024548,0.042987037,0.0054757944,0.43338242],"study_design_scores_gemma":[0.000016257476,0.00003714247,0.00007531137,0.000007655439,0.000004555343,0.000040479852,0.0000066255584,0.9885344,0.0059211343,0.0041469242,0.0012001239,0.000009423015],"about_ca_topic_score_codex":0.0021670475,"about_ca_topic_score_gemma":0.0025007236,"teacher_disagreement_score":0.0033875706,"about_ca_system_score_codex":0.00056794356,"about_ca_system_score_gemma":0.00066097104,"threshold_uncertainty_score":0.0113325715},"labels":[],"label_agreement":null},{"id":"W2133278758","doi":"10.1109/tsa.2004.825668","title":"Speaker Adaptation Using an Eigenphone Basis","year":2004,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Speech and Audio Processing","field":"Computer Science","cited_by":17,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Computer Research Institute of Montréal","funders":"","keywords":"Speech recognition; Adaptation (eye); Computer science; Basis (linear algebra); Mathematics; Psychology","score_opus":0.037026132462567876,"score_gpt":0.2671019762390837,"score_spread":0.23007584377651585,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2133278758","genre_codex":"methods","genre_gemma":"methods","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"methods","genre_consensus":"methods","domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0029872605,0.00013309877,0.9956145,0.00004526156,0.00006593317,0.000015771093,0.000036957124,0.00044115956,0.0006600612],"genre_scores_gemma":[0.09107227,0.00045683995,0.90014386,0.00018584206,0.0001657333,0.00015046004,0.0004242467,0.0003609159,0.00703974],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99908125,0.00026269793,0.00003265952,0.00021695027,0.0003451435,0.00006132636],"domain_scores_gemma":[0.99937016,0.00022924572,0.00003767497,0.000148595,0.0001846278,0.00002970829],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008481022,0.0006041368,0.0007883361,0.0007241071,0.00043527028,0.0006917498,0.0010566778,0.00095257245,0.0036235822],"category_scores_gemma":[0.0022405686,0.00053674326,0.0010820745,0.00075545243,0.0003566417,0.0013147522,0.001129973,0.0017026827,0.0030632478],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0001836099,0.00014008378,0.001233666,0.00010651923,0.00019268044,0.00013092882,0.00022260447,0.10156188,0.05580735,0.026996316,0.004553371,0.8088709],"study_design_scores_gemma":[0.000009913311,0.00003992098,0.00087114854,0.000012577923,0.00003226868,0.0001629297,0.000022060087,0.97018224,0.010691017,0.011696851,0.006240435,0.000038656926],"about_ca_topic_score_codex":0.0019996073,"about_ca_topic_score_gemma":0.0038866873,"teacher_disagreement_score":0.0036235822,"about_ca_system_score_codex":0.00024892617,"about_ca_system_score_gemma":0.0005750708,"threshold_uncertainty_score":0.012122095},"labels":[],"label_agreement":null},{"id":"W2135433537","doi":"10.1109/89.928919","title":"A maximum a posteriori approach to speaker adaptation using the trended hidden Markov model","year":2001,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Hidden Markov model; Maximum a posteriori estimation; Polynomial; Computer science; Speech recognition; Adaptation (eye); Security token; A priori and a posteriori; Gaussian; Pattern recognition (psychology); Mathematics; Algorithm; Artificial intelligence; Statistics; Maximum likelihood","score_opus":0.05320708971129079,"score_gpt":0.2664378542114752,"score_spread":0.2132307645001844,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2135433537","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0017952505,0.00007611202,0.9975604,0.000032405624,0.000012419389,0.000013621066,0.000022273058,0.00024645063,0.000241023],"genre_scores_gemma":[0.12157443,0.00044514058,0.8738481,0.00008404018,0.000112339316,0.0002177667,0.0003144291,0.00026678867,0.0031370774],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9991116,0.00037455797,0.000043666638,0.00021390778,0.00021036302,0.000045847642],"domain_scores_gemma":[0.99854904,0.0009732594,0.0000692964,0.00014755146,0.00023882044,0.000022109354],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001832936,0.00082522415,0.0007835683,0.0006302447,0.0004692294,0.0006953855,0.0011948369,0.0009646951,0.0017788529],"category_scores_gemma":[0.0053097066,0.0007137236,0.0011845253,0.0006449051,0.00048532026,0.0010826988,0.0009029773,0.0018183213,0.001095833],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0002645403,0.00008555743,0.00085790013,0.00019237523,0.00028357303,0.00017030958,0.00028817632,0.5486665,0.02383363,0.019499479,0.002393232,0.40346482],"study_design_scores_gemma":[0.000007903114,0.000039630522,0.000298491,0.000008579917,0.000025467878,0.000050061284,0.000012846285,0.986062,0.0032990212,0.008931574,0.0012445427,0.00001992438],"about_ca_topic_score_codex":0.003462002,"about_ca_topic_score_gemma":0.0041118856,"teacher_disagreement_score":0.003462002,"about_ca_system_score_codex":0.00039859972,"about_ca_system_score_gemma":0.0009808929,"threshold_uncertainty_score":0.009693563},"labels":[],"label_agreement":null},{"id":"W2135817141","doi":"10.1109/tsa.2004.833008","title":"Time-Delay Estimation via Linear Interpolation and Cross Correlation","year":2004,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Speech and Audio Processing","field":"Computer Science","cited_by":243,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Institut National de la Recherche Scientifique; Université du Québec à Montréal","funders":"","keywords":"Reverberation; Multilateration; Cross-correlation; Computer science; Microphone; Multipath propagation; Algorithm; Noise (video); Interpolation (computer graphics); SIGNAL (programming language); Linear interpolation; Speech recognition; Acoustics; Mathematics; Artificial intelligence; Telecommunications; Pattern recognition (psychology)","score_opus":0.00977493263216831,"score_gpt":0.25927492469456587,"score_spread":0.24949999206239756,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2135817141","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.006660045,0.00017413378,0.9922536,0.000025460413,0.000022788747,0.000011291451,0.000021617967,0.0003429293,0.00048811393],"genre_scores_gemma":[0.19442436,0.0005513509,0.80080485,0.000051083924,0.00006615481,0.00007365204,0.00029391586,0.00018358033,0.0035510731],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.99946016,0.00011411252,0.000027661305,0.00013205547,0.000216607,0.00004945629],"domain_scores_gemma":[0.99872833,0.00058504357,0.00015314686,0.00014878393,0.00034976986,0.00003495176],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00093919685,0.0008023282,0.00053411955,0.0011305306,0.00032241715,0.00069620117,0.0008653183,0.000630002,0.0017355954],"category_scores_gemma":[0.0035524163,0.0003548874,0.00063336204,0.0014605604,0.0003904032,0.0009381155,0.0009964828,0.0011231172,0.00078209874],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021351635,0.000090869675,0.002076574,0.00016765278,0.00010393955,0.00019648451,0.00014785262,0.4632329,0.026672488,0.016669923,0.0021216087,0.4883062],"study_design_scores_gemma":[0.000005577347,0.000033180648,0.00038471076,0.000008234362,0.000010412237,0.000083551495,0.000009689393,0.99041486,0.006056963,0.0014028143,0.0015739076,0.000016053762],"about_ca_topic_score_codex":0.0055816504,"about_ca_topic_score_gemma":0.0048756255,"teacher_disagreement_score":0.0055816504,"about_ca_system_score_codex":0.0004401126,"about_ca_system_score_gemma":0.0011435105,"threshold_uncertainty_score":0.011098325},"labels":[],"label_agreement":null},{"id":"W2136836257","doi":"10.1109/89.979381","title":"A robust compensation strategy for extraneous acoustic variations in spontaneous speech recognition","year":2002,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Speech Recognition and Synthesis","field":"Computer Science","cited_by":15,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Hidden Markov model; Speech recognition; Computer science; Word error rate; Pattern recognition (psychology); Pronunciation; Bayesian probability; Artificial intelligence","score_opus":0.07417596438202625,"score_gpt":0.2557498206924044,"score_spread":0.18157385631037812,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2136836257","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.02829794,0.000222077,0.9698738,0.00005421996,0.00004176222,0.00003644418,0.000046987814,0.0010216312,0.00040522622],"genre_scores_gemma":[0.47212312,0.00029438292,0.52360886,0.0001648824,0.00008954896,0.0002146589,0.00048849825,0.00019485917,0.002821241],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993579,0.00012852876,0.000046691195,0.00016814578,0.0002495802,0.000049187176],"domain_scores_gemma":[0.999342,0.0001654252,0.000092217284,0.00017236295,0.00019410028,0.00003385319],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00062812126,0.0007484098,0.0006161236,0.0002981406,0.00021130533,0.0003997089,0.0009519604,0.0006276482,0.0009952143],"category_scores_gemma":[0.0020170875,0.00028591402,0.0003636464,0.00026058836,0.000331465,0.0006888742,0.00053277693,0.0005069792,0.0008416689],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00030059216,0.00015763464,0.0009156337,0.0001176882,0.00008503897,0.0002106334,0.00012162557,0.030743632,0.3855463,0.0023944415,0.0014215958,0.5779852],"study_design_scores_gemma":[0.00005522085,0.0004662129,0.004329623,0.000014672537,0.00008455502,0.0011128393,0.000049068567,0.75903577,0.22741462,0.0021390002,0.005210404,0.00008802263],"about_ca_topic_score_codex":0.0010156455,"about_ca_topic_score_gemma":0.0013481709,"teacher_disagreement_score":0.0010156455,"about_ca_system_score_codex":0.00015328327,"about_ca_system_score_gemma":0.00043741113,"threshold_uncertainty_score":0.0033293366},"labels":[],"label_agreement":null},{"id":"W2137996830","doi":"10.1109/tsa.2005.851941","title":"A blind channel identification-based two-stage approach to separation and dereverberation of speech signals in a reverberant environment","year":2005,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Blind Source Separation Techniques","field":"Computer Science","cited_by":103,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Université du Québec","funders":"","keywords":"Reverberation; Blind signal separation; MIMO; Robustness (evolution); Computer science; Speech recognition; Source separation; Interference (communication); Channel (broadcasting); Signal processing; Acoustics; Algorithm; Telecommunications; Physics","score_opus":0.025129042935960907,"score_gpt":0.28828980738897314,"score_spread":0.26316076445301223,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2137996830","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0048504393,0.00008336382,0.9941993,0.000027156571,0.000019381403,0.000017225719,0.000012829606,0.00020089599,0.00058951584],"genre_scores_gemma":[0.20793505,0.0002954559,0.7855135,0.00007487712,0.00005591882,0.0001001659,0.000090734735,0.000042629166,0.0058916626],"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","domain_scores_codex":[0.99970275,0.00006654617,0.000018460565,0.00008794169,0.000090973095,0.000033315937],"domain_scores_gemma":[0.99974304,0.00008651624,0.000025077828,0.000044431825,0.00008535814,0.000015587788],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00029582708,0.0005657201,0.00054353854,0.0004156168,0.00039576265,0.00042417314,0.0005578955,0.00070699287,0.0016091818],"category_scores_gemma":[0.0007435763,0.00030473995,0.0005218984,0.00030985835,0.00064161967,0.0006680832,0.0005340055,0.00067127554,0.000717083],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0005959478,0.00018677049,0.0007056157,0.00025580844,0.000095836185,0.00023979586,0.00026395536,0.28882048,0.1964595,0.03414405,0.0018532224,0.47637904],"study_design_scores_gemma":[0.0000292712,0.00015945565,0.00045676233,0.0000062882878,0.00002037512,0.00022096842,0.000020041949,0.95170134,0.03871097,0.0061340216,0.0025022696,0.000038296155],"about_ca_topic_score_codex":0.0016545289,"about_ca_topic_score_gemma":0.0028312518,"teacher_disagreement_score":0.0016545289,"about_ca_system_score_codex":0.00037696963,"about_ca_system_score_gemma":0.00079305476,"threshold_uncertainty_score":0.005383253},"labels":[],"label_agreement":null},{"id":"W2145103350","doi":"10.1109/89.966081","title":"Linear prediction based packet loss concealment algorithm for PCM coded speech","year":2001,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Advanced Data Compression Techniques","field":"Computer Science","cited_by":83,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Nortel (Canada)","funders":"","keywords":"Linear prediction; Computer science; Speech coding; Algorithm; Packet loss; Speech recognition; Network packet; Voice activity detection; Linear predictive coding; Frame (networking); Pulse-code modulation; Speech processing; SIGNAL (programming language); Speech enhancement; Residual; Artificial intelligence; Noise reduction; Telecommunications; Computer network","score_opus":0.023244735467878783,"score_gpt":0.29098343852523606,"score_spread":0.26773870305735725,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2145103350","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.010162221,0.00024058277,0.98836833,0.000056820027,0.000028420109,0.000031688993,0.000016860966,0.0007654197,0.00032975562],"genre_scores_gemma":[0.18232834,0.000391269,0.8131051,0.00007793192,0.000042164265,0.000108543565,0.00013297223,0.000065738815,0.00374789],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995735,0.000070968694,0.000030955063,0.000064643864,0.00022787259,0.000032069103],"domain_scores_gemma":[0.99911004,0.000333417,0.00011531565,0.00009789895,0.00031957764,0.000023788607],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0009189594,0.00047183846,0.0004139044,0.00055870914,0.0003766887,0.0005698758,0.000785697,0.00062822993,0.0013974434],"category_scores_gemma":[0.002094222,0.00020595439,0.0002172008,0.000477722,0.00033539106,0.00090264704,0.000481672,0.0009063433,0.000801254],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.0006244583,0.00008891188,0.0007267376,0.000082857354,0.000030974854,0.00008495649,0.00013292847,0.089028426,0.04818764,0.0049558515,0.0018660296,0.8541902],"study_design_scores_gemma":[0.00003889014,0.000121408484,0.0004605179,0.0000095767955,0.000017725626,0.00012408769,0.00001705538,0.95169306,0.044470288,0.0010261588,0.0020046257,0.000016620508],"about_ca_topic_score_codex":0.002379912,"about_ca_topic_score_gemma":0.0023318713,"teacher_disagreement_score":0.002379912,"about_ca_system_score_codex":0.00052556355,"about_ca_system_score_gemma":0.0008097667,"threshold_uncertainty_score":0.004859984},"labels":[],"label_agreement":null},{"id":"W2146814186","doi":"10.1109/tsa.2005.851945","title":"A variable step-size pre-filter-bank adaptive algorithm","year":2005,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Advanced Adaptive Filtering Techniques","field":"Engineering","cited_by":4,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"Queen's University","funders":"","keywords":"Algorithm; Adaptive filter; Mathematics; Filter (signal processing); Computational complexity theory; Convergence (economics); Autocorrelation; Filter bank; Noise (video); Adaptive algorithm; A priori and a posteriori; Computer science; Statistics; Artificial intelligence","score_opus":0.01163436432685181,"score_gpt":0.23488334796582935,"score_spread":0.22324898363897755,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2146814186","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.0021716023,0.00008205829,0.9965389,0.000041472365,0.00003912574,0.0000352061,0.000012563205,0.00030558158,0.0007734385],"genre_scores_gemma":[0.12165151,0.00027026093,0.8706009,0.0001760155,0.00006882186,0.00029115216,0.00016262961,0.00008849265,0.0066901967],"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9995012,0.00007819359,0.000032467433,0.00013118536,0.00021726989,0.00003960006],"domain_scores_gemma":[0.99957234,0.00015271614,0.000032146778,0.000053087737,0.00017103917,0.000018664825],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00047313666,0.0005192061,0.0006356748,0.00050344015,0.00035034423,0.00067947013,0.0012720168,0.0010061973,0.003979674],"category_scores_gemma":[0.0013643712,0.00035578935,0.00047183756,0.00056461064,0.00042629038,0.0007216192,0.000618038,0.001080521,0.001317902],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00021885033,0.00009364938,0.0006434541,0.0001464943,0.000071687085,0.000104463164,0.00008745205,0.16575097,0.042878706,0.0153518785,0.004327428,0.770325],"study_design_scores_gemma":[0.000049239643,0.00010161267,0.00034942318,0.00001332267,0.000017474216,0.00014744357,0.000009504373,0.9815325,0.01015624,0.0020621158,0.005541936,0.000019149184],"about_ca_topic_score_codex":0.002422122,"about_ca_topic_score_gemma":0.0023824638,"teacher_disagreement_score":0.003979674,"about_ca_system_score_codex":0.00041539298,"about_ca_system_score_gemma":0.0012998391,"threshold_uncertainty_score":0.013313353},"labels":[],"label_agreement":null},{"id":"W2170959418","doi":"10.1109/tsa.2003.814411","title":"Quantization of lsf parameters using a trellis modeling","year":2003,"lang":"en","type":"article","venue":"IEEE Transactions on Speech and Audio Processing","topic":"Advanced Data Compression Techniques","field":"Computer Science","cited_by":10,"is_retracted":false,"has_abstract":true,"route_ca_aff":true,"route_ca_fund":false,"route_ca_venue":false,"route_about_ca":false,"ca_institutions":"University of Waterloo","funders":"","keywords":"Trellis quantization; Vector quantization; Algorithm; Quantization (signal processing); Speech recognition; Mathematics; Computer science; Trellis (graph); Speech coding; Decoding methods; Artificial intelligence; Image processing","score_opus":0.04364163151637969,"score_gpt":0.2915569645516571,"score_spread":0.2479153330352774,"validation_status":"score_only:v0-immature-baseline","prediction":{"id":"W2170959418","genre_codex":"methods","genre_gemma":"empirical","domain_codex":null,"domain_gemma":null,"model_version":"metacan-v3-hybrid-931329e0061c","genre_candidate":"empirical","genre_consensus":null,"domain_candidate":null,"domain_consensus":null,"prediction_status":"machine_predicted_unvalidated","genre_scores_codex":[0.003657015,0.00017821703,0.99476177,0.000050278297,0.000023794333,0.00003593749,0.00006859493,0.00033785607,0.0008866209],"genre_scores_gemma":[0.30335018,0.0008751864,0.68973005,0.00010803476,0.00007610349,0.0002562997,0.0005188342,0.00016574313,0.0049196198],"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","domain_scores_codex":[0.9993748,0.00015707937,0.000049215083,0.00008748638,0.00028328132,0.00004809166],"domain_scores_gemma":[0.9991886,0.00033445153,0.0000954241,0.00012144528,0.00024429918,0.000015735675],"candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0007036287,0.0005230044,0.000488507,0.0005304921,0.00034753926,0.0007858685,0.000771919,0.0006105533,0.0026486875],"category_scores_gemma":[0.0020208142,0.00030635696,0.0005949419,0.0007198848,0.00046473747,0.0010869286,0.00028550596,0.00093400764,0.0010424986],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_system_candidate":false,"about_ca_system_consensus":false,"study_design_scores_codex":[0.00017640651,0.00004998904,0.000412565,0.00014972211,0.000044830915,0.00010338657,0.00011107372,0.71452796,0.065877676,0.05966205,0.0023756314,0.15650868],"study_design_scores_gemma":[0.000008463563,0.000054882476,0.00009123177,0.000010936506,0.000008615039,0.00005265874,0.00000656242,0.98023903,0.013364778,0.003892994,0.0022544416,0.0000154642],"about_ca_topic_score_codex":0.004621288,"about_ca_topic_score_gemma":0.005157181,"teacher_disagreement_score":0.004621288,"about_ca_system_score_codex":0.0009451137,"about_ca_system_score_gemma":0.0009916887,"threshold_uncertainty_score":0.009188771},"labels":[],"label_agreement":null}]}