{"id":"https://openalex.org/W3140501190","doi":"https://doi.org/10.1109/taslp.2021.3069080","title":"Bayesian Learning of LF-MMI Trained Time Delay Neural Networks for Speech Recognition","display_name":"Bayesian Learning of LF-MMI Trained Time Delay Neural Networks for Speech Recognition","publication_year":2021,"publication_date":"2021-01-01","ids":{"openalex":"https://openalex.org/W3140501190","doi":"https://doi.org/10.1109/taslp.2021.3069080","mag":"3140501190"},"language":"en","primary_location":{"id":"doi:10.1109/taslp.2021.3069080","is_oa":false,"landing_page_url":"https://doi.org/10.1109/taslp.2021.3069080","pdf_url":null,"source":{"id":"https://openalex.org/S4210169297","display_name":"IEEE/ACM Transactions on Audio Speech and Language Processing","issn_l":"2329-9290","issn":["2329-9290","2329-9304"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE/ACM Transactions on Audio, Speech, and Language Processing","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5045355404","display_name":"Shoukang Hu","orcid":"https://orcid.org/0000-0002-3345-6923"},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Shoukang Hu","raw_affiliation_strings":["Department of Systems Engineering and Engineering Management, Chinese University of Hong Kong, Hong Kong SAR, China"],"raw_orcid":"https://orcid.org/0000-0002-3345-6923","affiliations":[{"raw_affiliation_string":"Department of Systems Engineering and Engineering Management, Chinese University of Hong Kong, Hong Kong SAR, China","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113476082","display_name":"Xurong Xie","orcid":"https://orcid.org/0009-0007-6729-2879"},"institutions":[{"id":"https://openalex.org/I4210145761","display_name":"Shenzhen Institutes of Advanced Technology","ror":"https://ror.org/04gh4er46","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210145761"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xurong Xie","raw_affiliation_strings":["Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China","institution_ids":["https://openalex.org/I4210145761"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5066070154","display_name":"Shansong Liu","orcid":"https://orcid.org/0000-0001-6202-5615"},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Shansong Liu","raw_affiliation_strings":["Department of Systems Engineering and Engineering Management, Chinese University of Hong Kong, Hong Kong SAR, China"],"raw_orcid":"https://orcid.org/0000-0001-6202-5615","affiliations":[{"raw_affiliation_string":"Department of Systems Engineering and Engineering Management, Chinese University of Hong Kong, Hong Kong SAR, China","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5004643540","display_name":"Jianwei Yu","orcid":"https://orcid.org/0000-0002-2449-1436"},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Jianwei Yu","raw_affiliation_strings":["Department of Systems Engineering and Engineering Management, Chinese University of Hong Kong, Hong Kong SAR, China"],"raw_orcid":"https://orcid.org/0000-0002-2449-1436","affiliations":[{"raw_affiliation_string":"Department of Systems Engineering and Engineering Management, Chinese University of Hong Kong, Hong Kong SAR, China","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100341722","display_name":"Zi Ye","orcid":"https://orcid.org/0000-0003-1002-0315"},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Zi Ye","raw_affiliation_strings":["Department of Systems Engineering and Engineering Management, Chinese University of Hong Kong, Hong Kong SAR, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Systems Engineering and Engineering Management, Chinese University of Hong Kong, Hong Kong SAR, China","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5003895235","display_name":"Mengzhe Geng","orcid":"https://orcid.org/0000-0002-7886-439X"},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Mengzhe Geng","raw_affiliation_strings":["Department of Systems Engineering and Engineering Management, Chinese University of Hong Kong, Hong Kong SAR, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Systems Engineering and Engineering Management, Chinese University of Hong Kong, Hong Kong SAR, China","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5037109470","display_name":"Xunying Liu","orcid":"https://orcid.org/0000-0001-6725-1160"},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Xunying Liu","raw_affiliation_strings":["Department of Systems Engineering and Engineering Management, Chinese University of Hong Kong, Hong Kong SAR, China"],"raw_orcid":"https://orcid.org/0000-0001-6725-1160","affiliations":[{"raw_affiliation_string":"Department of Systems Engineering and Engineering Management, Chinese University of Hong Kong, Hong Kong SAR, China","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5019458385","display_name":"Helen Meng","orcid":"https://orcid.org/0000-0002-4427-3532"},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Helen Meng","raw_affiliation_strings":["Department of Systems Engineering and Engineering Management, Chinese University of Hong Kong, Hong Kong SAR, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Systems Engineering and Engineering Management, Chinese University of Hong Kong, Hong Kong SAR, China","institution_ids":["https://openalex.org/I177725633"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":1.6919,"has_fulltext":false,"cited_by_count":16,"citation_normalized_percentile":{"value":0.86499399,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":89,"max":98},"biblio":{"volume":"29","issue":null,"first_page":"1514","last_page":"1529"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.9998999834060669,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.9998999834060669,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10860","display_name":"Speech and Audio Processing","score":0.9997000098228455,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11309","display_name":"Music and Audio Processing","score":0.9984999895095825,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7633475065231323},{"id":"https://openalex.org/keywords/discriminative-model","display_name":"Discriminative model","score":0.6748874187469482},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"Speech recognition","score":0.6191238760948181},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.5214771032333374},{"id":"https://openalex.org/keywords/word-error-rate","display_name":"Word error rate","score":0.5208264589309692},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.5150487422943115},{"id":"https://openalex.org/keywords/hidden-markov-model","display_name":"Hidden Markov model","score":0.48125502467155457},{"id":"https://openalex.org/keywords/overfitting","display_name":"Overfitting","score":0.4613035321235657},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.4277235269546509},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.34751224517822266}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7633475065231323},{"id":"https://openalex.org/C97931131","wikidata":"https://www.wikidata.org/wiki/Q5282087","display_name":"Discriminative model","level":2,"score":0.6748874187469482},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.6191238760948181},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5214771032333374},{"id":"https://openalex.org/C40969351","wikidata":"https://www.wikidata.org/wiki/Q3516228","display_name":"Word error rate","level":2,"score":0.5208264589309692},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.5150487422943115},{"id":"https://openalex.org/C23224414","wikidata":"https://www.wikidata.org/wiki/Q176769","display_name":"Hidden Markov model","level":2,"score":0.48125502467155457},{"id":"https://openalex.org/C22019652","wikidata":"https://www.wikidata.org/wiki/Q331309","display_name":"Overfitting","level":3,"score":0.4613035321235657},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4277235269546509},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.34751224517822266}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/taslp.2021.3069080","is_oa":false,"landing_page_url":"https://doi.org/10.1109/taslp.2021.3069080","pdf_url":null,"source":{"id":"https://openalex.org/S4210169297","display_name":"IEEE/ACM Transactions on Audio Speech and Language Processing","issn_l":"2329-9290","issn":["2329-9290","2329-9304"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE/ACM Transactions on Audio, Speech, and Language Processing","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.6000000238418579,"display_name":"Reduced inequalities","id":"https://metadata.un.org/sdg/10"}],"awards":[{"id":"https://openalex.org/G2819829024","display_name":null,"funder_award_id":"PiH/350/20","funder_id":"https://openalex.org/F4320326427","funder_display_name":"Innovation and Technology Fund"},{"id":"https://openalex.org/G6461416196","display_name":"Development of a Bilingual Speech Recognition System for the Elderly and Disabled Population with Speech Disorders","funder_award_id":"ITS/254/19","funder_id":"https://openalex.org/F4320326427","funder_display_name":"Innovation and Technology Fund"},{"id":"https://openalex.org/G7022449685","display_name":null,"funder_award_id":"MMT-p1-19","funder_id":"https://openalex.org/F4320326427","funder_display_name":"Innovation and Technology Fund"}],"funders":[{"id":"https://openalex.org/F4320326427","display_name":"Innovation and Technology Fund","ror":null}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":117,"referenced_works":["https://openalex.org/W14913773","https://openalex.org/W38527073","https://openalex.org/W126474493","https://openalex.org/W330298975","https://openalex.org/W582134693","https://openalex.org/W592244745","https://openalex.org/W1494198834","https://openalex.org/W1538131130","https://openalex.org/W1544131626","https://openalex.org/W1567512734","https://openalex.org/W1599512239","https://openalex.org/W1663973292","https://openalex.org/W1746819321","https://openalex.org/W1762484328","https://openalex.org/W1828163288","https://openalex.org/W1866206747","https://openalex.org/W1877570817","https://openalex.org/W1959608418","https://openalex.org/W1987238397","https://openalex.org/W2002342963","https://openalex.org/W2003123121","https://openalex.org/W2033565080","https://openalex.org/W2074037951","https://openalex.org/W2085628288","https://openalex.org/W2094147890","https://openalex.org/W2095705004","https://openalex.org/W2104760318","https://openalex.org/W2108677974","https://openalex.org/W2111051539","https://openalex.org/W2114016253","https://openalex.org/W2117671523","https://openalex.org/W2122272452","https://openalex.org/W2125234026","https://openalex.org/W2127141656","https://openalex.org/W2131342762","https://openalex.org/W2145249131","https://openalex.org/W2146871184","https://openalex.org/W2150769028","https://openalex.org/W2150907703","https://openalex.org/W2154887136","https://openalex.org/W2164273299","https://openalex.org/W2164411961","https://openalex.org/W2167277568","https://openalex.org/W2169163929","https://openalex.org/W2212676342","https://openalex.org/W2242047860","https://openalex.org/W2293009711","https://openalex.org/W2296705629","https://openalex.org/W2327501763","https://openalex.org/W2401396087","https://openalex.org/W2402146185","https://openalex.org/W2403195671","https://openalex.org/W2407080277","https://openalex.org/W2514741789","https://openalex.org/W2555706138","https://openalex.org/W2587848299","https://openalex.org/W2696967604","https://openalex.org/W2749679082","https://openalex.org/W2766678531","https://openalex.org/W2786459654","https://openalex.org/W2788853733","https://openalex.org/W2888867175","https://openalex.org/W2889191672","https://openalex.org/W2889282842","https://openalex.org/W2936481169","https://openalex.org/W2936774411","https://openalex.org/W2937814805","https://openalex.org/W2939173691","https://openalex.org/W2944255943","https://openalex.org/W2953561564","https://openalex.org/W2963340809","https://openalex.org/W2963711523","https://openalex.org/W2964052793","https://openalex.org/W2964059111","https://openalex.org/W2964182776","https://openalex.org/W2972581070","https://openalex.org/W2972692349","https://openalex.org/W2972942899","https://openalex.org/W2981857663","https://openalex.org/W3002595344","https://openalex.org/W3015726069","https://openalex.org/W3015766686","https://openalex.org/W3015899421","https://openalex.org/W3095311338","https://openalex.org/W3097747488","https://openalex.org/W3161873458","https://openalex.org/W4211049957","https://openalex.org/W4212863985","https://openalex.org/W4391602018","https://openalex.org/W6600588093","https://openalex.org/W6601546718","https://openalex.org/W6605168674","https://openalex.org/W6617145748","https://openalex.org/W6617744952","https://openalex.org/W6632100814","https://openalex.org/W6637956674","https://openalex.org/W6638749077","https://openalex.org/W6639216784","https://openalex.org/W6639822467","https://openalex.org/W6640963894","https://openalex.org/W6674330103","https://openalex.org/W6676315081","https://openalex.org/W6682398751","https://openalex.org/W6683993359","https://openalex.org/W6684488266","https://openalex.org/W6688648137","https://openalex.org/W6696982659","https://openalex.org/W6697336780","https://openalex.org/W6712930963","https://openalex.org/W6713250295","https://openalex.org/W6713280142","https://openalex.org/W6713762819","https://openalex.org/W6733596270","https://openalex.org/W6745256532","https://openalex.org/W6748587240","https://openalex.org/W6779264309","https://openalex.org/W6795108984"],"related_works":["https://openalex.org/W3040691452","https://openalex.org/W2136763963","https://openalex.org/W2109705048","https://openalex.org/W2940588515","https://openalex.org/W1909151225","https://openalex.org/W1987783679","https://openalex.org/W2160030256","https://openalex.org/W3151937861","https://openalex.org/W2162582511","https://openalex.org/W4245698648"],"abstract_inverted_index":{"Discriminative":[0],"training":[1,26,49,100],"techniques":[2,178],"define":[3],"state-of-the-art":[4,128,187],"performance":[5,22,188,207],"for":[6,43],"automatic":[7],"speech":[8,219,243],"recognition":[9,220],"systems.":[10,111],"However,":[11],"they":[12],"are":[13,62],"inherently":[14],"prone":[15],"to":[16,19,30,41,64,105,158,238,261],"overfitting,":[17],"leading":[18],"poor":[20],"generalization":[21],"when":[23],"using":[24,86,141,194,255],"limited":[25],"data.":[27],"In":[28],"order":[29],"address":[31],"this":[32,34],"issue,":[33],"paper":[35],"presents":[36],"a":[37,127,183,213,223,232,239],"full":[38],"Bayesian":[39,56,177,247],"framework":[40],"account":[42],"model":[44,65,152],"uncertainty":[45,67],"in":[46,98,156,182,203],"sequence":[47],"discriminative":[48],"of":[50,73,107,119,174],"factored":[51,138],"TDNN":[52,59,110,139,248],"acoustic":[53],"models.":[54],"Several":[55],"learning":[57,161],"based":[58,166],"variant":[60],"systems":[61,201,250],"proposed":[63,176,246],"the":[66,78,108,175,186,191,195,204,245,252],"over":[68,126],"weight":[69,257],"parameters":[70],"and":[71,101,154,169,199],"choices":[72],"hidden":[74,79,162],"activation":[75],"functions,":[76],"or":[77],"layer":[80],"outputs.":[81],"Efficient":[82],"variational":[83],"inference":[84],"approaches":[85],"as":[87,89],"few":[88],"one":[90],"single":[91],"parameter":[92],"sample":[93],"ensure":[94],"their":[95],"computational":[96],"cost":[97],"both":[99],"evaluation":[102],"time":[103],"comparable":[104],"that":[106],"baseline":[109,136,253],"Statistically":[112],"significant":[113],"word":[114],"error":[115],"rate":[116],"(WER)":[117],"reductions":[118],"0.4%-1.8%":[120],"absolute":[121,263],"(5%-11%":[122],"relative)":[123],"were":[124,209],"obtained":[125,189,211],"900":[129],"h":[130],"speed":[131],"perturbed":[132],"Switchboard":[133],"corpus":[134],"trained":[135,236],"LF-MMI":[137,249],"system":[140,237,254],"multiple":[142],"regularization":[143],"methods":[144],"including":[145],"F-smoothing,":[146],"L2":[147],"norm":[148],"penalty,":[149],"natural":[150],"gradient,":[151],"averaging":[153],"dropout,":[155],"addition":[157],"i-Vector":[159],"plus":[160],"unit":[163],"contribution":[164],"(LHUC)":[165],"speaker":[167],"adaptation":[168,227],"RNNLM":[170],"rescoring.":[171],"The":[172],"efficacy":[173],"is":[179],"further":[180],"demonstrated":[181],"comparison":[184],"against":[185],"on":[190,212],"same":[192],"task":[193,228],"most":[196],"recent":[197],"hybrid":[198],"end-to-end":[200],"reported":[202],"literature.":[205],"Consistent":[206],"improvements":[208],"also":[210],"450-h":[214],"HKUST":[215],"conversational":[216],"Mandarin":[217],"telephone":[218],"task.":[221],"On":[222],"third":[224],"cross":[225],"domain":[226],"requiring":[229],"rapidly":[230],"porting":[231],"1000-h":[233],"LibriSpeech":[234],"data":[235],"small":[240],"DementiaBank":[241],"elderly":[242],"corpus,":[244],"outperformed":[251],"direct":[256],"fine-tuning":[258],"by":[259],"up":[260],"2.5%":[262],"WER":[264],"reduction.":[265]},"counts_by_year":[{"year":2026,"cited_by_count":2},{"year":2025,"cited_by_count":1},{"year":2024,"cited_by_count":4},{"year":2023,"cited_by_count":2},{"year":2022,"cited_by_count":6},{"year":2021,"cited_by_count":1}],"updated_date":"2026-07-29T09:40:50.615796","created_date":"2025-10-10T00:00:00"}