{"id":"https://openalex.org/W3149629662","doi":"https://doi.org/10.1109/slt48900.2021.9383518","title":"RNN-T Models Fail to Generalize to Out-of-Domain Audio: Causes and Solutions","display_name":"RNN-T Models Fail to Generalize to Out-of-Domain Audio: Causes and Solutions","publication_year":2021,"publication_date":"2021-01-19","ids":{"openalex":"https://openalex.org/W3149629662","doi":"https://doi.org/10.1109/slt48900.2021.9383518","mag":"3149629662"},"language":"en","primary_location":{"id":"doi:10.1109/slt48900.2021.9383518","doi":"https://doi.org/10.1109/slt48900.2021.9383518","is_oa":false,"landing_page_url":"https://doi.org/10.1109/slt48900.2021.9383518","pdf_url":null,"source":{"id":"https://openalex.org/S4363605953","display_name":"IEEE Spoken Language Technology Workshop (SLT)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2021 IEEE Spoken Language Technology Workshop (SLT)","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","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/A5027763497","display_name":"Chung\u2010Cheng Chiu","orcid":"https://orcid.org/0000-0001-9729-4778","observed_orcids":["https://orcid.org/0000-0001-9729-4778"]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Chung-Cheng Chiu","raw_affiliation_strings":["D. E. Shaw Group, Google Inc"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"D. E. Shaw Group, Google Inc","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5000078382","display_name":"Arun Narayanan","orcid":"https://orcid.org/0009-0008-3325-8928","observed_orcids":["https://orcid.org/0009-0008-3325-8928"]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Arun Narayanan","raw_affiliation_strings":["D. E. Shaw Group, Google Inc"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"D. E. Shaw Group, Google Inc","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100750907","display_name":"Wei Han","orcid":"https://orcid.org/0000-0002-4201-9645","observed_orcids":["https://orcid.org/0000-0002-4201-9645"]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Wei Han","raw_affiliation_strings":["D. E. Shaw Group, Google Inc"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"D. E. Shaw Group, Google Inc","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5032640894","display_name":"Rohit Prabhavalkar","orcid":"https://orcid.org/0000-0001-5331-6058","observed_orcids":["https://orcid.org/0000-0001-5331-6058"]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Rohit Prabhavalkar","raw_affiliation_strings":["D. E. Shaw Group, Google Inc"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"D. E. Shaw Group, Google Inc","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100433648","display_name":"Yu Zhang","orcid":"https://orcid.org/0000-0002-9505-1833","observed_orcids":["https://orcid.org/0000-0002-9505-1833","https://orcid.org/0000-0003-4808-5560"]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yu Zhang","raw_affiliation_strings":["D. E. Shaw Group, Google Inc"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"D. E. Shaw Group, Google Inc","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5112445699","display_name":"Navdeep Jaitly","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Navdeep Jaitly","raw_affiliation_strings":["D. E. Shaw Group, Google Inc"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"D. E. Shaw Group, Google Inc","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5112901893","display_name":"Ruoming Pang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ruoming Pang","raw_affiliation_strings":["D. E. Shaw Group, Google Inc"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"D. E. Shaw Group, Google Inc","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5070513394","display_name":"Tara N. Sainath","orcid":"https://orcid.org/0000-0002-4126-6556","observed_orcids":["https://orcid.org/0000-0002-4126-6556"]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Tara N. Sainath","raw_affiliation_strings":["D. E. Shaw Group, Google Inc"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"D. E. Shaw Group, Google Inc","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5111455517","display_name":"Patrick Nguyen","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Patrick Nguyen","raw_affiliation_strings":["D. E. Shaw Group, Google Inc"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"D. E. Shaw Group, Google Inc","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103187717","display_name":"Liangliang Cao","orcid":"https://orcid.org/0000-0003-0900-1512","observed_orcids":["https://orcid.org/0000-0003-0900-1512"]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Liangliang Cao","raw_affiliation_strings":["D. E. Shaw Group, Google Inc"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"D. E. Shaw Group, Google Inc","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5010253402","display_name":"Yonghui Wu","orcid":"https://orcid.org/0000-0002-6780-6135","observed_orcids":["https://orcid.org/0000-0002-6780-6135"]},"institutions":[{"id":"https://openalex.org/I1291425158","display_name":"Google (United States)","ror":"https://ror.org/00njsd438","country_code":"US","type":"company","lineage":["https://openalex.org/I1291425158","https://openalex.org/I4210128969"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yonghui Wu","raw_affiliation_strings":["D. E. Shaw Group, Google Inc"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"D. E. Shaw Group, Google Inc","institution_ids":["https://openalex.org/I1291425158"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I1291425158"],"apc_list":null,"apc_paid":null,"fwci":2.8882,"has_fulltext":false,"cited_by_count":28,"citation_normalized_percentile":{"value":0.92792811,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":90,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"873","last_page":"880"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":1.0,"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":1.0,"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":0.0,"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":0.0,"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/recurrent-neural-network-transducer","display_name":"RNN-T","score":0.8679999709129333},{"id":"https://openalex.org/keywords/out-of-domain-generalization","display_name":"out-of-domain generalization","score":0.9010000228881836},{"id":"https://openalex.org/keywords/domain-mismatch","display_name":"domain mismatch","score":0.800000011920929},{"id":"https://openalex.org/keywords/streaming-asr","display_name":"streaming ASR","score":0.6069999933242798},{"id":"https://openalex.org/keywords/end-to-end-speech-recognition","display_name":"end-to-end speech recognition","score":0.7139999866485596},{"id":"https://openalex.org/keywords/regularization","display_name":"regularization","score":0.5320000052452087},{"id":"https://openalex.org/keywords/youtube-audio","display_name":"YouTube audio","score":0.609000027179718},{"id":"https://openalex.org/keywords/word-error-rate","display_name":"word error rate","score":0.6919999718666077}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7951408624649048},{"id":"https://openalex.org/C2776135515","wikidata":"https://www.wikidata.org/wiki/Q17143721","display_name":"Regularization (linguistics)","level":2,"score":0.7168409824371338},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6997758150100708},{"id":"https://openalex.org/C147168706","wikidata":"https://www.wikidata.org/wiki/Q1457734","display_name":"Recurrent neural network","level":3,"score":0.6662592887878418},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.6603758335113525},{"id":"https://openalex.org/C40969351","wikidata":"https://www.wikidata.org/wiki/Q3516228","display_name":"Word error rate","level":2,"score":0.578039288520813},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.5737457871437073},{"id":"https://openalex.org/C169903167","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Test set","level":2,"score":0.5669342875480652},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.46989771723747253},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.45201146602630615},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.45146963000297546},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.4349173307418823},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.42757177352905273},{"id":"https://openalex.org/C16910744","wikidata":"https://www.wikidata.org/wiki/Q7705759","display_name":"Test data","level":2,"score":0.4171258807182312},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.34022057056427},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.09055373072624207},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0},{"id":"https://openalex.org/C120665830","wikidata":"https://www.wikidata.org/wiki/Q14620","display_name":"Optics","level":1,"score":0.0},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.0},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/slt48900.2021.9383518","doi":"https://doi.org/10.1109/slt48900.2021.9383518","is_oa":false,"landing_page_url":"https://doi.org/10.1109/slt48900.2021.9383518","pdf_url":null,"source":{"id":"https://openalex.org/S4363605953","display_name":"IEEE Spoken Language Technology Workshop (SLT)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2021 IEEE Spoken Language Technology Workshop (SLT)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"sustainable_development_goals_aurora":[{"display_name":"Quality education","score":0.4699999988079071,"id":"https://metadata.un.org/sdg/4"}],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":33,"referenced_works":["https://openalex.org/W854541894","https://openalex.org/W1494198834","https://openalex.org/W1828163288","https://openalex.org/W1922655562","https://openalex.org/W2064675550","https://openalex.org/W2102113734","https://openalex.org/W2108677974","https://openalex.org/W2121879602","https://openalex.org/W2127141656","https://openalex.org/W2143612262","https://openalex.org/W2160815625","https://openalex.org/W2193413348","https://openalex.org/W2291513470","https://openalex.org/W2312434537","https://openalex.org/W2327501763","https://openalex.org/W2507771204","https://openalex.org/W2545177271","https://openalex.org/W2597757402","https://openalex.org/W2766219058","https://openalex.org/W2773781902","https://openalex.org/W2775304348","https://openalex.org/W2886246524","https://openalex.org/W2889129739","https://openalex.org/W2928941594","https://openalex.org/W2936774411","https://openalex.org/W2944255943","https://openalex.org/W2963414781","https://openalex.org/W2964272710","https://openalex.org/W2981532039","https://openalex.org/W3008898571","https://openalex.org/W3015194534","https://openalex.org/W3015995734","https://openalex.org/W6687566353"],"related_works":["https://openalex.org/W2160451571","https://openalex.org/W2495256954","https://openalex.org/W1566315437","https://openalex.org/W2259317772","https://openalex.org/W2594897229","https://openalex.org/W4221142855","https://openalex.org/W2151348424","https://openalex.org/W2050138804","https://openalex.org/W767271433","https://openalex.org/W4290708361"],"abstract_inverted_index":{"In":[0,60],"recent":[1],"years,":[2],"all-neural":[3],"end-to-end":[4,47,78],"approaches":[5],"have":[6],"obtained":[7],"state-of-the-art":[8],"results":[9,39],"on":[10,23,44,50,57,142,151,162,172],"several":[11],"challenging":[12],"automatic":[13],"speech":[14],"recognition":[15],"(ASR)":[16],"tasks.":[17],"However,":[18],"most":[19],"existing":[20],"works":[21],"focus":[22],"building":[24],"ASR":[25],"models":[26,48,79],"where":[27],"train":[28],"and":[29,70,101],"test":[30,110],"data":[31],"are":[32],"drawn":[33],"from":[34,132,155,174],"the":[35,65,113,124,137,146,152],"same":[36],"domain.":[37],"This":[38],"in":[40,80],"poor":[41],"generalization":[42,66,89],"characteristics":[43],"mismatched-domains:":[45],"e.g.,":[46],"trained":[49,118,141,161],"short":[51,143],"segments":[52,121],"perform":[53],"poorly":[54],"when":[55,112,136,160],"evaluated":[56],"longer":[58],"utterances.":[59],"this":[61],"work,":[62],"we":[63,164],"analyze":[64],"properties":[67],"of":[68,122],"streaming":[69,138],"non-streaming":[71,114],"recurrent":[72],"neural":[73],"network":[74],"transducer":[75],"(RNN-T)":[76],"based":[77],"order":[81],"to":[82,134,157,176],"identify":[83],"model":[84,116,140],"components":[85],"that":[86,166],"negatively":[87],"affect":[88],"performance.":[90],"We":[91],"propose":[92],"two":[93],"solutions:":[94],"combining":[95],"multiple":[96],"regularization":[97],"techniques":[98,148],"during":[99],"training,":[100],"using":[102],"dynamic":[103,167],"overlapping":[104,168],"inference.":[105],"On":[106],"a":[107],"long-form":[108],"YouTube":[109,153,173],"set,":[111],"RNN-T":[115,139],"is":[117],"with":[119],"shorter":[120],"data,":[123],"proposed":[125,147],"combination":[126],"improves":[127,170],"word":[128],"error":[129],"rate":[130],"(WER)":[131],"22.3%":[133],"14.8%;":[135],"Search":[144],"queries,":[145],"improve":[149],"WER":[150,171],"set":[154],"67.0%":[156],"25.3%.":[158],"Finally,":[159],"Librispeech,":[163],"find":[165],"inference":[169],"99.8%":[175],"33.0%.":[177]},"counts_by_year":[{"year":2025,"cited_by_count":1},{"year":2024,"cited_by_count":2},{"year":2023,"cited_by_count":11},{"year":2022,"cited_by_count":5},{"year":2021,"cited_by_count":7},{"year":2020,"cited_by_count":2}],"updated_date":"2026-10-16T09:59:22.235173","created_date":"2025-10-10T00:00:00"}