{"id":"https://openalex.org/W3047778367","doi":"https://doi.org/10.21437/interspeech.2020-1569","title":"Subword Regularization: An Analysis of Scalability and Generalization for End-to-End Automatic Speech Recognition","display_name":"Subword Regularization: An Analysis of Scalability and Generalization for End-to-End Automatic Speech Recognition","publication_year":2020,"publication_date":"2020-10-25","ids":{"openalex":"https://openalex.org/W3047778367","doi":"https://doi.org/10.21437/interspeech.2020-1569","mag":"3047778367"},"language":"en","primary_location":{"id":"doi:10.21437/interspeech.2020-1569","is_oa":false,"landing_page_url":"https://doi.org/10.21437/interspeech.2020-1569","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Interspeech 2020","raw_type":"proceedings-article"},"type":"preprint","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2008.04034","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5045428440","display_name":"Egor Lakomkin","orcid":null},"institutions":[{"id":"https://openalex.org/I159176309","display_name":"Universit\u00e4t Hamburg","ror":"https://ror.org/00g30e956","country_code":"DE","type":"education","lineage":["https://openalex.org/I159176309"]}],"countries":["DE"],"is_corresponding":true,"raw_author_name":"Egor Lakomkin","raw_affiliation_strings":["Universit\u00e4t Hamburg, Hamburg, Germany"],"affiliations":[{"raw_affiliation_string":"Universit\u00e4t Hamburg, Hamburg, Germany","institution_ids":["https://openalex.org/I159176309"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5006131416","display_name":"Jahn Heymann","orcid":null},"institutions":[{"id":"https://openalex.org/I1311688040","display_name":"Amazon (United States)","ror":"https://ror.org/04mv4n011","country_code":"US","type":"company","lineage":["https://openalex.org/I1311688040"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jahn Heymann","raw_affiliation_strings":["Amazon (United States), Seattle, United States"],"affiliations":[{"raw_affiliation_string":"Amazon (United States), Seattle, United States","institution_ids":["https://openalex.org/I1311688040"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5064043969","display_name":"Ilya Sklyar","orcid":null},"institutions":[{"id":"https://openalex.org/I1311688040","display_name":"Amazon (United States)","ror":"https://ror.org/04mv4n011","country_code":"US","type":"company","lineage":["https://openalex.org/I1311688040"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ilya Sklyar","raw_affiliation_strings":["Amazon (United States), Seattle, United States"],"affiliations":[{"raw_affiliation_string":"Amazon (United States), Seattle, United States","institution_ids":["https://openalex.org/I1311688040"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5077727156","display_name":"Simon Wiesler","orcid":null},"institutions":[{"id":"https://openalex.org/I1311688040","display_name":"Amazon (United States)","ror":"https://ror.org/04mv4n011","country_code":"US","type":"company","lineage":["https://openalex.org/I1311688040"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Simon Wiesler","raw_affiliation_strings":["Amazon (United States), Seattle, United States"],"affiliations":[{"raw_affiliation_string":"Amazon (United States), Seattle, United States","institution_ids":["https://openalex.org/I1311688040"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5045428440"],"corresponding_institution_ids":["https://openalex.org/I159176309"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.09136201,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"3600","last_page":"3604"},"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9998000264167786,"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/T10028","display_name":"Topic Modeling","score":0.9991999864578247,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7684017419815063},{"id":"https://openalex.org/keywords/regularization","display_name":"Regularization (linguistics)","score":0.6750351190567017},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"Speech recognition","score":0.6589290499687195},{"id":"https://openalex.org/keywords/segmentation","display_name":"Segmentation","score":0.6003775596618652},{"id":"https://openalex.org/keywords/vocabulary","display_name":"Vocabulary","score":0.5955593585968018},{"id":"https://openalex.org/keywords/word-error-rate","display_name":"Word error rate","score":0.538068413734436},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.5047386884689331},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.5042010545730591},{"id":"https://openalex.org/keywords/end-to-end-principle","display_name":"End-to-end principle","score":0.48778802156448364},{"id":"https://openalex.org/keywords/hidden-markov-model","display_name":"Hidden Markov model","score":0.43854427337646484},{"id":"https://openalex.org/keywords/deep-neural-networks","display_name":"Deep neural networks","score":0.41836318373680115},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.3507001996040344},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.33413374423980713},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.24586239457130432},{"id":"https://openalex.org/keywords/linguistics","display_name":"Linguistics","score":0.07472652196884155}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7684017419815063},{"id":"https://openalex.org/C2776135515","wikidata":"https://www.wikidata.org/wiki/Q17143721","display_name":"Regularization (linguistics)","level":2,"score":0.6750351190567017},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.6589290499687195},{"id":"https://openalex.org/C89600930","wikidata":"https://www.wikidata.org/wiki/Q1423946","display_name":"Segmentation","level":2,"score":0.6003775596618652},{"id":"https://openalex.org/C2777601683","wikidata":"https://www.wikidata.org/wiki/Q6499736","display_name":"Vocabulary","level":2,"score":0.5955593585968018},{"id":"https://openalex.org/C40969351","wikidata":"https://www.wikidata.org/wiki/Q3516228","display_name":"Word error rate","level":2,"score":0.538068413734436},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5047386884689331},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5042010545730591},{"id":"https://openalex.org/C74296488","wikidata":"https://www.wikidata.org/wiki/Q2527392","display_name":"End-to-end principle","level":2,"score":0.48778802156448364},{"id":"https://openalex.org/C23224414","wikidata":"https://www.wikidata.org/wiki/Q176769","display_name":"Hidden Markov model","level":2,"score":0.43854427337646484},{"id":"https://openalex.org/C2984842247","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep neural networks","level":3,"score":0.41836318373680115},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3507001996040344},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.33413374423980713},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.24586239457130432},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.07472652196884155},{"id":"https://openalex.org/C77088390","wikidata":"https://www.wikidata.org/wiki/Q8513","display_name":"Database","level":1,"score":0.0},{"id":"https://openalex.org/C138885662","wikidata":"https://www.wikidata.org/wiki/Q5891","display_name":"Philosophy","level":0,"score":0.0}],"mesh":[],"locations_count":4,"locations":[{"id":"doi:10.21437/interspeech.2020-1569","is_oa":false,"landing_page_url":"https://doi.org/10.21437/interspeech.2020-1569","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Interspeech 2020","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2008.04034","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2008.04034","pdf_url":"https://arxiv.org/pdf/2008.04034","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"mag:3047778367","is_oa":true,"landing_page_url":"http://export.arxiv.org/pdf/2008.04034","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"arXiv (Cornell University)","raw_type":null},{"id":"doi:10.48550/arxiv.2008.04034","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2008.04034","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2008.04034","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2008.04034","pdf_url":"https://arxiv.org/pdf/2008.04034","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[{"display_name":"Quality Education","score":0.6899999976158142,"id":"https://metadata.un.org/sdg/4"}],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W3047778367.pdf","grobid_xml":"https://content.openalex.org/works/W3047778367.grobid-xml"},"referenced_works_count":21,"referenced_works":["https://openalex.org/W1494198834","https://openalex.org/W1828163288","https://openalex.org/W2064675550","https://openalex.org/W2121879602","https://openalex.org/W2127141656","https://openalex.org/W2193413348","https://openalex.org/W2250303366","https://openalex.org/W2327501763","https://openalex.org/W2545177271","https://openalex.org/W2936774411","https://openalex.org/W2938348542","https://openalex.org/W2962760690","https://openalex.org/W2962784628","https://openalex.org/W2963414781","https://openalex.org/W2963979492","https://openalex.org/W2964121744","https://openalex.org/W2972630480","https://openalex.org/W2982413405","https://openalex.org/W3007227084","https://openalex.org/W3015995734","https://openalex.org/W3035207248"],"related_works":["https://openalex.org/W3095229326","https://openalex.org/W2965428162","https://openalex.org/W2074382434","https://openalex.org/W3162138321","https://openalex.org/W3182245104","https://openalex.org/W2178770890","https://openalex.org/W2964182350","https://openalex.org/W2737108017","https://openalex.org/W2950414763","https://openalex.org/W1585396714","https://openalex.org/W1494243555","https://openalex.org/W1516698649","https://openalex.org/W2284662682","https://openalex.org/W2402112010","https://openalex.org/W2182518606","https://openalex.org/W2137079242","https://openalex.org/W2139088233","https://openalex.org/W2890197052","https://openalex.org/W1966324591","https://openalex.org/W2736984005"],"abstract_inverted_index":{"Subwords":[0],"are":[1,63,68],"the":[2,14,21,29,72,110,114,130,136,139,174],"most":[3,73],"widely":[4],"used":[5],"output":[6],"units":[7,42],"in":[8,158],"end-to-end":[9,122],"speech":[10,35,94,123],"recognition.":[11],"They":[12],"combine":[13],"best":[15],"of":[16,23,113,138,152,168,176,181],"two":[17],"worlds":[18],"by":[19,37],"modeling":[20],"majority":[22],"frequent":[24],"words":[25,47,183],"directly":[26],"and":[27,58,93,184],"at":[28],"same":[30],"time":[31],"allow":[32],"open":[33],"vocabulary":[34],"recognition":[36,95,124,180],"backing":[38],"off":[39],"to":[40,45,54,98,165],"shorter":[41],"or":[43],"characters":[44],"construct":[46],"unseen":[48,182],"during":[49,83],"training.":[50],"However,":[51],"mapping":[52],"text":[53],"subwords":[55],"is":[56],"ambiguous":[57],"often":[59],"multiple":[60],"segmentation":[61,116],"variants":[62],"possible.":[64],"Yet,":[65],"many":[66],"systems":[67],"trained":[69],"using":[70],"only":[71],"likely":[74],"segmentation.":[75],"Recent":[76],"research":[77],"suggests":[78],"that":[79,145],"sampling":[80,117],"subword":[81,115,131,146,177],"segmentations":[82],"training":[84,140],"acts":[85],"as":[86],"a":[87,106,120,149,159,166],"regularizer":[88],"for":[89,119],"neural":[90],"machine":[91],"translation":[92],"models,":[96],"leading":[97],"performance":[99],"improvements.":[100],"In":[101,126],"this":[102],"work,":[103],"we":[104,128,172],"conduct":[105],"principled":[107],"investigation":[108],"on":[109,135,179,187],"regularizing":[111],"effect":[112,175],"method":[118],"streaming":[121],"task.":[125],"particular,":[127],"evaluate":[129],"regularization":[132,147,178],"contribution":[133],"depending":[134],"size":[137,167],"dataset.":[141],"Our":[142],"results":[143],"suggest":[144],"provides":[148],"consistent":[150],"improvement":[151],"(2-8%)":[153],"relative":[154],"word-error-rate":[155],"reduction,":[156],"even":[157],"large-scale":[160],"setting":[161],"with":[162],"datasets":[163],"up":[164],"20k":[169],"hours.":[170],"Further,":[171],"analyze":[173],"its":[185],"implications":[186],"beam":[188],"diversity.":[189]},"counts_by_year":[],"updated_date":"2026-03-20T23:20:44.827607","created_date":"2025-10-10T00:00:00"}