{"id":"https://openalex.org/W4210766398","doi":"https://doi.org/10.1109/asru51503.2021.9687921","title":"Comparative Study of Different Tokenization Strategies for Streaming End-to-End ASR","display_name":"Comparative Study of Different Tokenization Strategies for Streaming End-to-End ASR","publication_year":2021,"publication_date":"2021-12-13","ids":{"openalex":"https://openalex.org/W4210766398","doi":"https://doi.org/10.1109/asru51503.2021.9687921"},"language":"en","primary_location":{"id":"doi:10.1109/asru51503.2021.9687921","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asru51503.2021.9687921","pdf_url":null,"source":{"id":"https://openalex.org/S4363606113","display_name":"2021 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"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 Automatic Speech Recognition and Understanding Workshop (ASRU)","raw_type":"proceedings-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/A5072672307","display_name":"Sachin Singh","orcid":"https://orcid.org/0000-0003-2788-5462"},"institutions":[{"id":"https://openalex.org/I4210101778","display_name":"Samsung (United States)","ror":"https://ror.org/01bfbvm65","country_code":"US","type":"company","lineage":["https://openalex.org/I2250650973","https://openalex.org/I4210101778"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Sachin Singh","raw_affiliation_strings":["Samsung Research"],"affiliations":[{"raw_affiliation_string":"Samsung Research","institution_ids":["https://openalex.org/I4210101778"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5069263468","display_name":"Ashutosh Gupta","orcid":"https://orcid.org/0000-0002-3622-1740"},"institutions":[{"id":"https://openalex.org/I4210101778","display_name":"Samsung (United States)","ror":"https://ror.org/01bfbvm65","country_code":"US","type":"company","lineage":["https://openalex.org/I2250650973","https://openalex.org/I4210101778"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ashutosh Gupta","raw_affiliation_strings":["Samsung Research"],"affiliations":[{"raw_affiliation_string":"Samsung Research","institution_ids":["https://openalex.org/I4210101778"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5038271343","display_name":"Aman Maghan","orcid":null},"institutions":[{"id":"https://openalex.org/I4210101778","display_name":"Samsung (United States)","ror":"https://ror.org/01bfbvm65","country_code":"US","type":"company","lineage":["https://openalex.org/I2250650973","https://openalex.org/I4210101778"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Aman Maghan","raw_affiliation_strings":["Samsung Research"],"affiliations":[{"raw_affiliation_string":"Samsung Research","institution_ids":["https://openalex.org/I4210101778"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113125574","display_name":"Dhananjaya Gowda","orcid":null},"institutions":[{"id":"https://openalex.org/I4210101778","display_name":"Samsung (United States)","ror":"https://ror.org/01bfbvm65","country_code":"US","type":"company","lineage":["https://openalex.org/I2250650973","https://openalex.org/I4210101778"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Dhananjaya Gowda","raw_affiliation_strings":["Samsung Research"],"affiliations":[{"raw_affiliation_string":"Samsung Research","institution_ids":["https://openalex.org/I4210101778"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5079810318","display_name":"Shatrughan Singh","orcid":"https://orcid.org/0000-0003-3758-3214"},"institutions":[{"id":"https://openalex.org/I4210101778","display_name":"Samsung (United States)","ror":"https://ror.org/01bfbvm65","country_code":"US","type":"company","lineage":["https://openalex.org/I2250650973","https://openalex.org/I4210101778"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Shatrughan Singh","raw_affiliation_strings":["Samsung Research"],"affiliations":[{"raw_affiliation_string":"Samsung Research","institution_ids":["https://openalex.org/I4210101778"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100684422","display_name":"Chanwoo Kim","orcid":"https://orcid.org/0000-0003-0193-8167"},"institutions":[{"id":"https://openalex.org/I4210101778","display_name":"Samsung (United States)","ror":"https://ror.org/01bfbvm65","country_code":"US","type":"company","lineage":["https://openalex.org/I2250650973","https://openalex.org/I4210101778"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Chanwoo Kim","raw_affiliation_strings":["Samsung Research"],"affiliations":[{"raw_affiliation_string":"Samsung Research","institution_ids":["https://openalex.org/I4210101778"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5072672307"],"corresponding_institution_ids":["https://openalex.org/I4210101778"],"apc_list":null,"apc_paid":null,"fwci":0.2513,"has_fulltext":false,"cited_by_count":4,"citation_normalized_percentile":{"value":0.53897925,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":89,"max":96},"biblio":{"volume":null,"issue":null,"first_page":"388","last_page":"394"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","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/T10181","display_name":"Natural Language Processing Techniques","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/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/T10028","display_name":"Topic Modeling","score":0.9980999827384949,"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":0.8333042860031128},{"id":"https://openalex.org/keywords/byte","display_name":"Byte","score":0.7471380829811096},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.6795341372489929},{"id":"https://openalex.org/keywords/lexical-analysis","display_name":"Lexical analysis","score":0.6762136816978455},{"id":"https://openalex.org/keywords/vocabulary","display_name":"Vocabulary","score":0.6355468034744263},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.5740636587142944},{"id":"https://openalex.org/keywords/machine-translation","display_name":"Machine translation","score":0.5694018602371216},{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.5491072535514832},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.5217093825340271},{"id":"https://openalex.org/keywords/word","display_name":"Word (group theory)","score":0.5172089338302612},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"Speech recognition","score":0.5161124467849731},{"id":"https://openalex.org/keywords/character","display_name":"Character (mathematics)","score":0.5154796242713928},{"id":"https://openalex.org/keywords/word-error-rate","display_name":"Word error rate","score":0.4860899746417999},{"id":"https://openalex.org/keywords/linguistics","display_name":"Linguistics","score":0.24144643545150757},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.08427000045776367},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.08412262797355652}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8333042860031128},{"id":"https://openalex.org/C43364308","wikidata":"https://www.wikidata.org/wiki/Q8799","display_name":"Byte","level":2,"score":0.7471380829811096},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.6795341372489929},{"id":"https://openalex.org/C176982825","wikidata":"https://www.wikidata.org/wiki/Q835922","display_name":"Lexical analysis","level":2,"score":0.6762136816978455},{"id":"https://openalex.org/C2777601683","wikidata":"https://www.wikidata.org/wiki/Q6499736","display_name":"Vocabulary","level":2,"score":0.6355468034744263},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.5740636587142944},{"id":"https://openalex.org/C203005215","wikidata":"https://www.wikidata.org/wiki/Q79798","display_name":"Machine translation","level":2,"score":0.5694018602371216},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.5491072535514832},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5217093825340271},{"id":"https://openalex.org/C90805587","wikidata":"https://www.wikidata.org/wiki/Q10944557","display_name":"Word (group theory)","level":2,"score":0.5172089338302612},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.5161124467849731},{"id":"https://openalex.org/C2780861071","wikidata":"https://www.wikidata.org/wiki/Q1062934","display_name":"Character (mathematics)","level":2,"score":0.5154796242713928},{"id":"https://openalex.org/C40969351","wikidata":"https://www.wikidata.org/wiki/Q3516228","display_name":"Word error rate","level":2,"score":0.4860899746417999},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.24144643545150757},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.08427000045776367},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.08412262797355652},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.0},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"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":1,"locations":[{"id":"doi:10.1109/asru51503.2021.9687921","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asru51503.2021.9687921","pdf_url":null,"source":{"id":"https://openalex.org/S4363606113","display_name":"2021 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"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 Automatic Speech Recognition and Understanding Workshop (ASRU)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.7699999809265137,"display_name":"Quality Education","id":"https://metadata.un.org/sdg/4"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":33,"referenced_works":["https://openalex.org/W1494198834","https://openalex.org/W1522301498","https://openalex.org/W1810943226","https://openalex.org/W2064675550","https://openalex.org/W2095705004","https://openalex.org/W2102113734","https://openalex.org/W2799800213","https://openalex.org/W2936774411","https://openalex.org/W2962760690","https://openalex.org/W2962784628","https://openalex.org/W2964002616","https://openalex.org/W2971840980","https://openalex.org/W2973122799","https://openalex.org/W2991213871","https://openalex.org/W2997454826","https://openalex.org/W2998353611","https://openalex.org/W3007528493","https://openalex.org/W3008587939","https://openalex.org/W3026470569","https://openalex.org/W3095410713","https://openalex.org/W3095867028","https://openalex.org/W3096032230","https://openalex.org/W3124386064","https://openalex.org/W4287726212","https://openalex.org/W6631190155","https://openalex.org/W6674330103","https://openalex.org/W6675365184","https://openalex.org/W6756000814","https://openalex.org/W6770506093","https://openalex.org/W6777389638","https://openalex.org/W6780612146","https://openalex.org/W6784765489","https://openalex.org/W6789719078"],"related_works":["https://openalex.org/W2151197196","https://openalex.org/W2757141711","https://openalex.org/W4385571757","https://openalex.org/W2463396630","https://openalex.org/W4287760117","https://openalex.org/W3045818198","https://openalex.org/W3034697300","https://openalex.org/W4281478209","https://openalex.org/W1566315437","https://openalex.org/W4378498597"],"abstract_inverted_index":{"Most":[0],"End-to-End":[1],"Automatic":[2],"Speech":[3],"Recognition":[4],"(ASR)":[5],"models":[6,173,208],"use":[7],"character-based":[8],"vocabularies:":[9],"characters,":[10],"sub-words":[11,89],"(BPE),":[12],"or":[13],"words.":[14],"While":[15,182],"these":[16,30],"work":[17],"well":[18],"for":[19,85,97,138,146,178,183,238],"training":[20],"a":[21,32,65,175,184,214],"monolingual":[22,148,199],"model,":[23],"there":[24],"are":[25,91,101,136],"certain":[26],"limitations":[27],"when":[28],"ap-plying":[29],"to":[31,154,231],"multilingual":[33],"model.":[34],"Rare":[35],"characters":[36],"from":[37],"character-rich":[38,185],"languages":[39,152],"like":[40,187],"Korean":[41],"can":[42],"easily":[43],"result":[44],"in":[45,93,104],"large":[46],"vocabulary":[47,216],"size,":[48,130],"limiting":[49],"the":[50,56,77,105,117,132,206,218,228,233],"model's":[51],"compactness.":[52],"Repre-senting":[53],"text":[54,70],"at":[55,217],"level":[57],"of":[58,69,220],"bytes":[59],"has":[60],"also":[61],"been":[62],"proposed.":[63],"However,":[64],"byte":[66,207],"sequence":[67],"representation":[68,99],"is":[71],"often":[72],"much":[73],"longer,":[74],"which":[75,135],"increases":[76],"decoding":[78,133],"time":[79],"and":[80,131,149,157,159,162,171,180,192,200,213,236],"makes":[81],"it":[82],"computationally":[83],"expensive":[84],"on-device":[86,140],"use.":[87],"Byte-based":[88],"(BBPE)":[90],"proposed":[92],"neural":[94],"machine":[95],"translation":[96],"word":[98],"but":[100],"still":[102],"unexplored":[103],"ASR":[106],"domain.":[107],"In":[108,204],"this":[109],"work,":[110],"we":[111,189,226],"conduct":[112],"an":[113,139],"empirical":[114],"study":[115],"comparing":[116],"above":[118],"three":[119,123],"tokenization":[120],"strategies":[121],"across":[122],"metrics:":[124],"Word":[125],"Error":[126],"Rate":[127],"(WER),":[128],"model":[129,211],"time,":[134],"critical":[137],"ASR.":[141],"We":[142],"did":[143],"extensive":[144],"experiments":[145],"both":[147],"bilingual,":[150],"with":[151,197],"belonging":[153],"same":[155],"(English":[156,161],"Spanish)":[158],"different":[160],"Korean)":[163],"language":[164,186],"families.":[165],"Our":[166],"exper-iments":[167],"show":[168],"that":[169],"BBPE":[170,198,229],"BPE":[172],"yield":[174],"similar":[176],"WER":[177,195],"English":[179],"Spanish.":[181],"Korean,":[188],"get":[190],"26%":[191],"14%":[193],"relative":[194],"improvement":[196],"bilingual":[201],"models,":[202],"respectively.":[203],"contrast,":[205],"trade-off":[209],"small":[210],"size":[212],"fixed":[215],"cost":[219],"high":[221],"xRT.":[222],"Among":[223],"all":[224],"three,":[225],"found":[227],"strategy":[230],"be":[232],"most":[234,239],"flexible":[235],"optimal":[237],"cases.":[240]},"counts_by_year":[{"year":2025,"cited_by_count":2},{"year":2024,"cited_by_count":1},{"year":2023,"cited_by_count":1}],"updated_date":"2026-01-22T23:29:09.771500","created_date":"2025-10-10T00:00:00"}