{"id":"https://openalex.org/W4226302796","doi":"https://doi.org/10.1109/asru51503.2021.9688302","title":"Distilling Knowledge from Ensembles of Acoustic Models for Joint CTC-Attention End-to-End Speech Recognition","display_name":"Distilling Knowledge from Ensembles of Acoustic Models for Joint CTC-Attention End-to-End Speech Recognition","publication_year":2021,"publication_date":"2021-12-13","ids":{"openalex":"https://openalex.org/W4226302796","doi":"https://doi.org/10.1109/asru51503.2021.9688302"},"language":"en","primary_location":{"id":"doi:10.1109/asru51503.2021.9688302","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asru51503.2021.9688302","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":"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/A5100778706","display_name":"Yan Gao","orcid":"https://orcid.org/0000-0002-6852-2038"},"institutions":[{"id":"https://openalex.org/I241749","display_name":"University of Cambridge","ror":"https://ror.org/013meh722","country_code":"GB","type":"education","lineage":["https://openalex.org/I241749"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Yan Gao","raw_affiliation_strings":["University of Cambridge,United Kingdom","University of Cambridge, United Kingdom"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Cambridge,United Kingdom","institution_ids":["https://openalex.org/I241749"]},{"raw_affiliation_string":"University of Cambridge, United Kingdom","institution_ids":["https://openalex.org/I241749"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5089505434","display_name":"Titouan Parcollet","orcid":"https://orcid.org/0000-0003-0672-1346"},"institutions":[{"id":"https://openalex.org/I198415970","display_name":"Universit\u00e9 d'Avignon et des Pays de Vaucluse","ror":"https://ror.org/00mfpxb84","country_code":"FR","type":"education","lineage":["https://openalex.org/I198415970"]},{"id":"https://openalex.org/I241749","display_name":"University of Cambridge","ror":"https://ror.org/013meh722","country_code":"GB","type":"education","lineage":["https://openalex.org/I241749"]},{"id":"https://openalex.org/I4210119991","display_name":"Laboratoire Informatique d'Avignon","ror":"https://ror.org/02n399288","country_code":"FR","type":"facility","lineage":["https://openalex.org/I198415970","https://openalex.org/I4210119991"]}],"countries":["FR","GB"],"is_corresponding":false,"raw_author_name":"Titouan Parcollet","raw_affiliation_strings":["Avignon University,France","Avignon University, France","University of Cambridge, United Kingdom"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Avignon University,France","institution_ids":["https://openalex.org/I198415970","https://openalex.org/I4210119991"]},{"raw_affiliation_string":"Avignon University, France","institution_ids":["https://openalex.org/I198415970"]},{"raw_affiliation_string":"University of Cambridge, United Kingdom","institution_ids":["https://openalex.org/I241749"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5045638679","display_name":"Nicholas D. Lane","orcid":"https://orcid.org/0000-0002-2728-8273"},"institutions":[{"id":"https://openalex.org/I241749","display_name":"University of Cambridge","ror":"https://ror.org/013meh722","country_code":"GB","type":"education","lineage":["https://openalex.org/I241749"]},{"id":"https://openalex.org/I4210117523","display_name":"Samsung (United Kingdom)","ror":"https://ror.org/01w6gjq94","country_code":"GB","type":"company","lineage":["https://openalex.org/I2250650973","https://openalex.org/I4210117523"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Nicholas D. Lane","raw_affiliation_strings":["University of Cambridge,United Kingdom","Samsung AI, Cambridge, United-Kingdom","University of Cambridge, United Kingdom"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Cambridge,United Kingdom","institution_ids":["https://openalex.org/I241749"]},{"raw_affiliation_string":"Samsung AI, Cambridge, United-Kingdom","institution_ids":["https://openalex.org/I4210117523"]},{"raw_affiliation_string":"University of Cambridge, United Kingdom","institution_ids":["https://openalex.org/I241749"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":14,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"138","last_page":"145"},"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/T11309","display_name":"Music and Audio Processing","score":0.9991000294685364,"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":0.9987000226974487,"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":0.816175103187561},{"id":"https://openalex.org/keywords/timit","display_name":"TIMIT","score":0.7856762409210205},{"id":"https://openalex.org/keywords/end-to-end-principle","display_name":"End-to-end principle","score":0.6715224981307983},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"Speech recognition","score":0.6233925223350525},{"id":"https://openalex.org/keywords/distillation","display_name":"Distillation","score":0.5858501195907593},{"id":"https://openalex.org/keywords/word-error-rate","display_name":"Word error rate","score":0.5190263986587524},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.48492246866226196},{"id":"https://openalex.org/keywords/metric","display_name":"Metric (unit)","score":0.44329822063446045},{"id":"https://openalex.org/keywords/selection","display_name":"Selection (genetic algorithm)","score":0.4272860884666443},{"id":"https://openalex.org/keywords/acoustic-model","display_name":"Acoustic model","score":0.42036744952201843},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.40592676401138306},{"id":"https://openalex.org/keywords/hidden-markov-model","display_name":"Hidden Markov model","score":0.2884657382965088},{"id":"https://openalex.org/keywords/speech-processing","display_name":"Speech processing","score":0.23184514045715332},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.06847524642944336}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.816175103187561},{"id":"https://openalex.org/C2778724510","wikidata":"https://www.wikidata.org/wiki/Q7670405","display_name":"TIMIT","level":3,"score":0.7856762409210205},{"id":"https://openalex.org/C74296488","wikidata":"https://www.wikidata.org/wiki/Q2527392","display_name":"End-to-end principle","level":2,"score":0.6715224981307983},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.6233925223350525},{"id":"https://openalex.org/C204030448","wikidata":"https://www.wikidata.org/wiki/Q101017","display_name":"Distillation","level":2,"score":0.5858501195907593},{"id":"https://openalex.org/C40969351","wikidata":"https://www.wikidata.org/wiki/Q3516228","display_name":"Word error rate","level":2,"score":0.5190263986587524},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.48492246866226196},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.44329822063446045},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.4272860884666443},{"id":"https://openalex.org/C155635449","wikidata":"https://www.wikidata.org/wiki/Q4674699","display_name":"Acoustic model","level":3,"score":0.42036744952201843},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.40592676401138306},{"id":"https://openalex.org/C23224414","wikidata":"https://www.wikidata.org/wiki/Q176769","display_name":"Hidden Markov model","level":2,"score":0.2884657382965088},{"id":"https://openalex.org/C61328038","wikidata":"https://www.wikidata.org/wiki/Q3358061","display_name":"Speech processing","level":2,"score":0.23184514045715332},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.06847524642944336},{"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/C21547014","wikidata":"https://www.wikidata.org/wiki/Q1423657","display_name":"Operations management","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1109/asru51503.2021.9688302","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asru51503.2021.9688302","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"},{"id":"pmh:oai:HAL:hal-03601273v1","is_oa":false,"landing_page_url":"https://hal.science/hal-03601273","pdf_url":null,"source":{"id":"https://openalex.org/S4306402512","display_name":"HAL (Le Centre pour la Communication Scientifique Directe)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I1294671590","host_organization_name":"Centre National de la Recherche Scientifique","host_organization_lineage":["https://openalex.org/I1294671590"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"2021 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU), Dec 2021, Cartagena, Colombia. pp.138-145, ⟨10.1109/ASRU51503.2021.9688302⟩","raw_type":"info:eu-repo/semantics/conferenceObject"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.8199999928474426,"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":38,"referenced_works":["https://openalex.org/W1494198834","https://openalex.org/W1821462560","https://openalex.org/W2102113734","https://openalex.org/W2508418541","https://openalex.org/W2526425061","https://openalex.org/W2633884958","https://openalex.org/W2747909401","https://openalex.org/W2750784772","https://openalex.org/W2888930363","https://openalex.org/W2890153145","https://openalex.org/W2892008152","https://openalex.org/W2938974662","https://openalex.org/W2941286327","https://openalex.org/W2944255943","https://openalex.org/W2962826786","https://openalex.org/W2963209464","https://openalex.org/W2963514416","https://openalex.org/W2963723401","https://openalex.org/W2963736842","https://openalex.org/W2964203871","https://openalex.org/W2964227577","https://openalex.org/W2973040747","https://openalex.org/W3021654819","https://openalex.org/W3025567392","https://openalex.org/W3102424508","https://openalex.org/W3103005696","https://openalex.org/W3105966348","https://openalex.org/W3167533889","https://openalex.org/W4248018214","https://openalex.org/W4298159529","https://openalex.org/W6638523607","https://openalex.org/W6675365184","https://openalex.org/W6732837357","https://openalex.org/W6733321041","https://openalex.org/W6743188669","https://openalex.org/W6745722055","https://openalex.org/W6748163181","https://openalex.org/W6768086466"],"related_works":["https://openalex.org/W2994894110","https://openalex.org/W4377862891","https://openalex.org/W2005708641","https://openalex.org/W2050397613","https://openalex.org/W2080458333","https://openalex.org/W1995562189","https://openalex.org/W172689163","https://openalex.org/W1507177964","https://openalex.org/W2611200336","https://openalex.org/W4288040805"],"abstract_inverted_index":{"Knowledge":[0],"distillation":[1,31,55,68],"has":[2,37],"been":[3],"widely":[4],"used":[5],"to":[6,57,76,82],"compress":[7],"existing":[8],"deep":[9],"learning":[10],"models":[11,36],"while":[12],"preserving":[13],"the":[14,23,78,83,91,102,105,142],"performance":[15],"on":[16,90,121,141],"a":[17,116],"wide":[18],"range":[19],"of":[20,26,34,53,118],"applications.":[21],"In":[22,46,94,134],"specific":[24],"context":[25],"Automatic":[27],"Speech":[28],"Recognition":[29],"(ASR),":[30],"from":[32],"ensembles":[33],"acoustic":[35],"recently":[38],"shown":[39],"promising":[40],"results":[41],"in":[42],"increasing":[43],"recognition":[44],"performance.":[45],"this":[47,95],"paper,":[48],"we":[49,97],"propose":[50],"an":[51],"extension":[52],"multi-teacher":[54],"methods":[56],"joint":[58],"CTC-attention":[59],"end-to-end":[60],"ASR":[61],"systems.":[62],"We":[63,111],"also":[64],"introduce":[65],"three":[66],"novel":[67],"strategies.":[69],"The":[70],"core":[71],"intuition":[72],"behind":[73],"them":[74],"is":[75],"integrate":[77],"error":[79,137],"rate":[80],"metric":[81,107],"teacher":[84],"selection":[85,117],"rather":[86],"than":[87],"solely":[88],"focusing":[89],"observed":[92],"losses.":[93],"way,":[96],"directly":[98],"distill":[99],"and":[100,128,147],"optimize":[101],"student":[103],"toward":[104],"relevant":[106],"for":[108],"speech":[109],"recognition.":[110],"evaluate":[112],"these":[113],"strategies":[114],"under":[115],"training":[119],"procedures":[120],"different":[122],"datasets":[123],"(TIMIT,":[124],"Librispeech,":[125],"Common":[126,143],"Voice)":[127],"various":[129],"languages":[130],"(English,":[131],"French,":[132,145],"Italian).":[133],"particular,":[135],"state-of-the-art":[136],"rates":[138],"are":[139],"reported":[140],"Voice":[144],"Italian":[146],"TIMIT":[148],"datasets.":[149]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":2},{"year":2024,"cited_by_count":4},{"year":2023,"cited_by_count":5},{"year":2022,"cited_by_count":1},{"year":2021,"cited_by_count":1}],"updated_date":"2026-07-27T08:26:11.824852","created_date":"2025-10-10T00:00:00"}