{"id":"https://openalex.org/W4375850708","doi":"https://doi.org/10.1109/icassp49357.2023.10096418","title":"Comparison of Soft and Hard Target RNN-T Distillation for Large-Scale ASR","display_name":"Comparison of Soft and Hard Target RNN-T Distillation for Large-Scale ASR","publication_year":2023,"publication_date":"2023-05-05","ids":{"openalex":"https://openalex.org/W4375850708","doi":"https://doi.org/10.1109/icassp49357.2023.10096418"},"language":"en","primary_location":{"id":"doi:10.1109/icassp49357.2023.10096418","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49357.2023.10096418","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","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/A5014088530","display_name":"Dongseong Hwang","orcid":null},"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":true,"raw_author_name":"Dongseong Hwang","raw_affiliation_strings":["Google LLC,USA","Google LLC, USA"],"affiliations":[{"raw_affiliation_string":"Google LLC,USA","institution_ids":["https://openalex.org/I1291425158"]},{"raw_affiliation_string":"Google LLC, USA","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5032798707","display_name":"Khe Chai Sim","orcid":"https://orcid.org/0000-0002-0866-2223"},"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":"Khe Chai Sim","raw_affiliation_strings":["Google LLC,USA","Google LLC, USA"],"affiliations":[{"raw_affiliation_string":"Google LLC,USA","institution_ids":["https://openalex.org/I1291425158"]},{"raw_affiliation_string":"Google LLC, USA","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5092336773","display_name":"Yu Zhang","orcid":null},"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":["Google LLC,USA","Google LLC, USA"],"affiliations":[{"raw_affiliation_string":"Google LLC,USA","institution_ids":["https://openalex.org/I1291425158"]},{"raw_affiliation_string":"Google LLC, USA","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5032931723","display_name":"Trevor Strohman","orcid":null},"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":"Trevor Strohman","raw_affiliation_strings":["Google LLC,USA","Google LLC, USA"],"affiliations":[{"raw_affiliation_string":"Google LLC,USA","institution_ids":["https://openalex.org/I1291425158"]},{"raw_affiliation_string":"Google LLC, USA","institution_ids":["https://openalex.org/I1291425158"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5014088530"],"corresponding_institution_ids":["https://openalex.org/I1291425158"],"apc_list":null,"apc_paid":null,"fwci":1.3724,"has_fulltext":false,"cited_by_count":7,"citation_normalized_percentile":{"value":0.82830963,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":90,"max":97},"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"5"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.9997000098228455,"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":0.9997000098228455,"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":0.9959999918937683,"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":0.9896000027656555,"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/distillation","display_name":"Distillation","score":0.8217745423316956},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.729400634765625},{"id":"https://openalex.org/keywords/scale","display_name":"Scale (ratio)","score":0.5939602851867676},{"id":"https://openalex.org/keywords/transfer-of-learning","display_name":"Transfer of learning","score":0.5449317097663879},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.4833485782146454},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4831104278564453},{"id":"https://openalex.org/keywords/word-error-rate","display_name":"Word error rate","score":0.44494086503982544},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"Speech recognition","score":0.33207982778549194}],"concepts":[{"id":"https://openalex.org/C204030448","wikidata":"https://www.wikidata.org/wiki/Q101017","display_name":"Distillation","level":2,"score":0.8217745423316956},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.729400634765625},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.5939602851867676},{"id":"https://openalex.org/C150899416","wikidata":"https://www.wikidata.org/wiki/Q1820378","display_name":"Transfer of learning","level":2,"score":0.5449317097663879},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4833485782146454},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4831104278564453},{"id":"https://openalex.org/C40969351","wikidata":"https://www.wikidata.org/wiki/Q3516228","display_name":"Word error rate","level":2,"score":0.44494086503982544},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.33207982778549194},{"id":"https://openalex.org/C185592680","wikidata":"https://www.wikidata.org/wiki/Q2329","display_name":"Chemistry","level":0,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0},{"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/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icassp49357.2023.10096418","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49357.2023.10096418","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Quality Education","score":0.8799999952316284,"id":"https://metadata.un.org/sdg/4"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":35,"referenced_works":["https://openalex.org/W1821462560","https://openalex.org/W1828163288","https://openalex.org/W2962369866","https://openalex.org/W2963122170","https://openalex.org/W2973049979","https://openalex.org/W3008181812","https://openalex.org/W3015194534","https://openalex.org/W3026041220","https://openalex.org/W3028069987","https://openalex.org/W3034368386","https://openalex.org/W3035160371","https://openalex.org/W3036601975","https://openalex.org/W3093579165","https://openalex.org/W3097777922","https://openalex.org/W3138154797","https://openalex.org/W3160525311","https://openalex.org/W3160766462","https://openalex.org/W3162649911","https://openalex.org/W3196855257","https://openalex.org/W3204696009","https://openalex.org/W3213305054","https://openalex.org/W4221161761","https://openalex.org/W4225529283","https://openalex.org/W4226033575","https://openalex.org/W4226406363","https://openalex.org/W6638523607","https://openalex.org/W6638749077","https://openalex.org/W6765939562","https://openalex.org/W6769722537","https://openalex.org/W6777642411","https://openalex.org/W6780218876","https://openalex.org/W6784614252","https://openalex.org/W6804146617","https://openalex.org/W6810673746","https://openalex.org/W6810745558"],"related_works":["https://openalex.org/W2309273277","https://openalex.org/W2061937230","https://openalex.org/W1769849273","https://openalex.org/W1574295218","https://openalex.org/W113247760","https://openalex.org/W2547793174","https://openalex.org/W2070212102","https://openalex.org/W2544241817","https://openalex.org/W2158882055","https://openalex.org/W2914603036"],"abstract_inverted_index":{"Knowledge":[0],"distillation":[1,31,55,104],"is":[2,37],"an":[3],"effective":[4,81],"machine":[5],"learning":[6],"technique":[7],"to":[8,15,56,152],"transfer":[9],"knowledge":[10,30],"from":[11],"a":[12,16,116,124],"teacher":[13,84,93,113,151],"model":[14,118],"smaller":[17],"student":[18,86],"model,":[19,35],"especially":[20],"with":[21,119,142],"unlabeled":[22],"data.":[23],"In":[24],"this":[25],"paper,":[26],"we":[27,48,122],"focus":[28],"on":[29,61,131,136],"for":[32],"the":[33,62,83,99],"RNN-T":[34,59],"which":[36],"widely":[38],"used":[39],"in":[40,107],"state-of-the-art":[41],"(SoTA)":[42],"automatic":[43],"speech":[44],"recognition":[45],"(ASR).":[46],"Specifically,":[47],"compared":[49],"using":[50,138],"soft":[51,102,143],"and":[52,68,85,94],"hard":[53,77],"target":[54,103,144],"train":[57],"large-scale":[58],"models":[60],"LibriSpeech/LibriLight":[63],"public":[64],"dataset":[65],"(60k":[66],"hours)":[67],"our":[69,149],"in-house":[70],"data":[71,155],"(600k":[72],"hours).":[73],"We":[74],"found":[75],"that":[76],"targets":[78],"are":[79],"more":[80],"when":[82],"have":[87],"different":[88],"architecture,":[89],"such":[90],"as":[91],"large":[92,112,117],"small":[95],"streaming":[96],"student.":[97],"On":[98],"other":[100],"hand,":[101],"works":[105],"better":[106],"self-training":[108],"scenario":[109],"like":[110],"iterative":[111],"training.":[114],"For":[115],"0.6B":[120],"weights,":[121],"achieve":[123],"new":[125,154],"SoTA":[126],"word":[127],"error":[128],"rate":[129],"(WER)":[130],"LibriSpeech":[132],"(8%":[133],"relative":[134],"improvement":[135],"dev-other)":[137],"Noisy":[139],"Student":[140],"Training":[141],"distillation.":[145],"It":[146],"also":[147],"allows":[148],"production":[150],"adapt":[153],"domain":[156],"continuously.":[157]},"counts_by_year":[{"year":2025,"cited_by_count":3},{"year":2024,"cited_by_count":1},{"year":2023,"cited_by_count":3}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}