{"id":"https://openalex.org/W3197813307","doi":"https://doi.org/10.1109/asru51503.2021.9687874","title":"Efficient Conformer: Progressive Downsampling and Grouped Attention for Automatic Speech Recognition","display_name":"Efficient Conformer: Progressive Downsampling and Grouped Attention for Automatic Speech Recognition","publication_year":2021,"publication_date":"2021-12-13","ids":{"openalex":"https://openalex.org/W3197813307","doi":"https://doi.org/10.1109/asru51503.2021.9687874","mag":"3197813307"},"language":"en","primary_location":{"id":"doi:10.1109/asru51503.2021.9687874","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asru51503.2021.9687874","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/A5057004182","display_name":"Maxime Burchi","orcid":null},"institutions":[{"id":"https://openalex.org/I19370010","display_name":"Orange (France)","ror":"https://ror.org/035j0tq82","country_code":"FR","type":"company","lineage":["https://openalex.org/I19370010"]}],"countries":["FR"],"is_corresponding":true,"raw_author_name":"Maxime Burchi","raw_affiliation_strings":["Orange Labs,Cesson-S&#x00E9;vign&#x00E9;,France"],"affiliations":[{"raw_affiliation_string":"Orange Labs,Cesson-S&#x00E9;vign&#x00E9;,France","institution_ids":["https://openalex.org/I19370010"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5019268455","display_name":"Valentin Vielzeuf","orcid":"https://orcid.org/0000-0002-2987-2510"},"institutions":[{"id":"https://openalex.org/I19370010","display_name":"Orange (France)","ror":"https://ror.org/035j0tq82","country_code":"FR","type":"company","lineage":["https://openalex.org/I19370010"]}],"countries":["FR"],"is_corresponding":false,"raw_author_name":"Valentin Vielzeuf","raw_affiliation_strings":["Orange Labs,Cesson-S&#x00E9;vign&#x00E9;,France"],"affiliations":[{"raw_affiliation_string":"Orange Labs,Cesson-S&#x00E9;vign&#x00E9;,France","institution_ids":["https://openalex.org/I19370010"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":["https://openalex.org/A5057004182"],"corresponding_institution_ids":["https://openalex.org/I19370010"],"apc_list":null,"apc_paid":null,"fwci":6.5989,"has_fulltext":false,"cited_by_count":72,"citation_normalized_percentile":{"value":0.97624988,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":99,"max":100},"biblio":{"volume":null,"issue":null,"first_page":"8","last_page":"15"},"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/T11309","display_name":"Music and Audio Processing","score":0.998199999332428,"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/T10860","display_name":"Speech and Audio Processing","score":0.9962999820709229,"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/upsampling","display_name":"Upsampling","score":0.7906812429428101},{"id":"https://openalex.org/keywords/conformational-isomerism","display_name":"Conformational isomerism","score":0.6434528827667236},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.58203125},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4934665262699127},{"id":"https://openalex.org/keywords/encoder","display_name":"Encoder","score":0.4451926648616791},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.324165403842926},{"id":"https://openalex.org/keywords/image","display_name":"Image (mathematics)","score":0.16178825497627258},{"id":"https://openalex.org/keywords/operating-system","display_name":"Operating system","score":0.0835985541343689}],"concepts":[{"id":"https://openalex.org/C110384440","wikidata":"https://www.wikidata.org/wiki/Q1143270","display_name":"Upsampling","level":3,"score":0.7906812429428101},{"id":"https://openalex.org/C18705241","wikidata":"https://www.wikidata.org/wiki/Q1128023","display_name":"Conformational isomerism","level":3,"score":0.6434528827667236},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.58203125},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4934665262699127},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.4451926648616791},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.324165403842926},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.16178825497627258},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.0835985541343689},{"id":"https://openalex.org/C178790620","wikidata":"https://www.wikidata.org/wiki/Q11351","display_name":"Organic chemistry","level":1,"score":0.0},{"id":"https://openalex.org/C185592680","wikidata":"https://www.wikidata.org/wiki/Q2329","display_name":"Chemistry","level":0,"score":0.0},{"id":"https://openalex.org/C32909587","wikidata":"https://www.wikidata.org/wiki/Q11369","display_name":"Molecule","level":2,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/asru51503.2021.9687874","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asru51503.2021.9687874","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":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":68,"referenced_works":["https://openalex.org/W1494198834","https://openalex.org/W1522301498","https://openalex.org/W1828163288","https://openalex.org/W1855892484","https://openalex.org/W1922655562","https://openalex.org/W1964175594","https://openalex.org/W2079667486","https://openalex.org/W2127141656","https://openalex.org/W2134800885","https://openalex.org/W2143612262","https://openalex.org/W2194775991","https://openalex.org/W2294370754","https://openalex.org/W2294543795","https://openalex.org/W2520160253","https://openalex.org/W2567070169","https://openalex.org/W2752782242","https://openalex.org/W2892009249","https://openalex.org/W2949718784","https://openalex.org/W2955425717","https://openalex.org/W2962760690","https://openalex.org/W2963163009","https://openalex.org/W2963250244","https://openalex.org/W2963403868","https://openalex.org/W2963414781","https://openalex.org/W2963928591","https://openalex.org/W2963970792","https://openalex.org/W2964110616","https://openalex.org/W2964121744","https://openalex.org/W2970971581","https://openalex.org/W2972818416","https://openalex.org/W2973215447","https://openalex.org/W2981413347","https://openalex.org/W2991526275","https://openalex.org/W3007328579","https://openalex.org/W3015537910","https://openalex.org/W3015995734","https://openalex.org/W3016010032","https://openalex.org/W3032960712","https://openalex.org/W3095173472","https://openalex.org/W3097777922","https://openalex.org/W3120633509","https://openalex.org/W3141854550","https://openalex.org/W3152519008","https://openalex.org/W3163793923","https://openalex.org/W3168612151","https://openalex.org/W3172509117","https://openalex.org/W3197917733","https://openalex.org/W4295312788","https://openalex.org/W4298395628","https://openalex.org/W4385245566","https://openalex.org/W4394650770","https://openalex.org/W6631190155","https://openalex.org/W6638749077","https://openalex.org/W6639156005","https://openalex.org/W6640090968","https://openalex.org/W6679855610","https://openalex.org/W6727336983","https://openalex.org/W6731370813","https://openalex.org/W6739901393","https://openalex.org/W6747398299","https://openalex.org/W6762718338","https://openalex.org/W6763367864","https://openalex.org/W6763509872","https://openalex.org/W6766978945","https://openalex.org/W6771571646","https://openalex.org/W6792376965","https://openalex.org/W6793959113","https://openalex.org/W6796463219"],"related_works":["https://openalex.org/W2023036309","https://openalex.org/W1972863456","https://openalex.org/W2089371831","https://openalex.org/W2062399876","https://openalex.org/W2052399476","https://openalex.org/W2810163160","https://openalex.org/W3083970380","https://openalex.org/W2037964119","https://openalex.org/W4220905048","https://openalex.org/W4385077663"],"abstract_inverted_index":{"The":[0],"recently":[1],"proposed":[2,141],"Conformer":[3,34,60,206],"architecture":[4,35,142],"has":[5],"shown":[6],"state-of-the-art":[7],"performances":[8,145],"in":[9],"Automatic":[10],"Speech":[11],"Recog-nition":[12],"by":[13],"combining":[14],"convolution":[15],"with":[16,37,127,146,172],"attention":[17,66,75],"to":[18,31,43,58,73,81,152,213],"model":[19,159,169,177],"both":[20],"local":[21],"and":[22,62,97,129,149,170,210],"global":[23,116],"dependencies.":[24],"In":[25],"this":[26],"paper,":[27],"we":[28,50],"study":[29],"how":[30],"reduce":[32,74],"the":[33,59,108,124,136,140,153,179],"complexity":[36,76],"a":[38,44,64,115,167,194],"limited":[39],"computing":[40,138],"budget,":[41,139],"leading":[42],"more":[45],"efficient":[46],"archi-tecture":[47],"design":[48],"that":[49,134],"call":[51],"Efficient":[52],"Conformer.":[53,154],"We":[54,105,132],"introduce":[55],"progressive":[56],"downsampling":[57,117],"encoder":[61],"propose":[63],"novel":[65],"mechanism":[67],"named":[68],"grouped":[69],"attention,":[70],"allowing":[71],"us":[72],"from":[77],"<tex":[78,82,88,94,101],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[79,83,89,95,102,186,189],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">$O(n^{2}d)$</tex>":[80],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">$O(n^{2}d/g)$</tex>":[84],"for":[85],"sequence":[86],"length":[87],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">$n$</tex>":[90],",":[91],"hidden":[92],"dimension":[93],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">$d$</tex>":[96],"group":[98],"size":[99],"parameter":[100],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">$g$</tex>":[103],".":[104],"also":[106],"experiment":[107],"use":[109],"of":[110,163],"strided":[111],"multi-head":[112],"self-attention":[113],"as":[114],"operation.":[118],"Our":[119,155],"experiments":[120],"are":[121],"performed":[122],"on":[123,178,193],"LibriSpeech":[125],"dataset":[126],"CTC":[128,158,205],"RNN-Transducer":[130],"losses.":[131],"show":[133],"within":[135],"same":[137],"achieves":[143,160],"better":[144],"faster":[147,202,212],"training":[148],"decoding":[150],"compared":[151],"13M":[156],"parameters":[157],"competitive":[161],"WERs":[162],"3.6%/9.0%":[164],"without":[165],"using":[166],"language":[168,176],"2.7%/6.7%":[171],"an":[173],"external":[174],"n-gram":[175],"test-clean/test-other":[180],"sets":[181],"while":[182],"being":[183],"29%":[184],"<sup":[185,188],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">1</sup>":[187,190],"Inference":[191],"time":[192],"single":[195],"Intel":[196],"Core":[197],"i9-9940X":[198],"3.3GHz":[199],"CPU":[200],"thread":[201],"than":[203],"our":[204],"baseline":[207],"at":[208],"inference":[209],"36%":[211],"train.":[214]},"counts_by_year":[{"year":2026,"cited_by_count":2},{"year":2025,"cited_by_count":18},{"year":2024,"cited_by_count":24},{"year":2023,"cited_by_count":17},{"year":2022,"cited_by_count":11}],"updated_date":"2026-03-28T08:17:26.163206","created_date":"2025-10-10T00:00:00"}