{"id":"https://openalex.org/W4296070216","doi":"https://doi.org/10.21437/interspeech.2022-10613","title":"A Scalable Model Specialization Framework for Training and Inference using Submodels and its Application to Speech Model Personalization","display_name":"A Scalable Model Specialization Framework for Training and Inference using Submodels and its Application to Speech Model Personalization","publication_year":2022,"publication_date":"2022-09-16","ids":{"openalex":"https://openalex.org/W4296070216","doi":"https://doi.org/10.21437/interspeech.2022-10613"},"language":"en","primary_location":{"id":"doi:10.21437/interspeech.2022-10613","is_oa":false,"landing_page_url":"https://doi.org/10.21437/interspeech.2022-10613","pdf_url":null,"source":{"id":"https://openalex.org/S4363604309","display_name":"Interspeech 2022","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":"Interspeech 2022","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/A5055592425","display_name":"Fadi Biadsy","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":"Fadi Biadsy","raw_affiliation_strings":["Google LLC"],"affiliations":[{"raw_affiliation_string":"Google LLC","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5025973834","display_name":"Youzheng Chen","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":"Youzheng Chen","raw_affiliation_strings":["Google LLC"],"affiliations":[{"raw_affiliation_string":"Google LLC","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100461895","display_name":"Xia Zhang","orcid":"https://orcid.org/0000-0001-7005-4387"},"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":"Xia Zhang","raw_affiliation_strings":["Google LLC"],"affiliations":[{"raw_affiliation_string":"Google LLC","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5070398306","display_name":"Oleg Rybakov\u200e","orcid":"https://orcid.org/0000-0003-4805-3083"},"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":"Oleg Rybakov","raw_affiliation_strings":["Google LLC"],"affiliations":[{"raw_affiliation_string":"Google LLC","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5024913801","display_name":"Andrew E. Rosenberg","orcid":"https://orcid.org/0000-0002-4536-8824"},"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":"Andrew Rosenberg","raw_affiliation_strings":["Google LLC"],"affiliations":[{"raw_affiliation_string":"Google LLC","institution_ids":["https://openalex.org/I1291425158"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5038592676","display_name":"Pedro Moreno","orcid":"https://orcid.org/0000-0003-3745-5845"},"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":"Pedro Moreno","raw_affiliation_strings":["Google LLC"],"affiliations":[{"raw_affiliation_string":"Google LLC","institution_ids":["https://openalex.org/I1291425158"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5055592425"],"corresponding_institution_ids":["https://openalex.org/I1291425158"],"apc_list":null,"apc_paid":null,"fwci":0.5194,"has_fulltext":false,"cited_by_count":5,"citation_normalized_percentile":{"value":0.63429218,"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":"5125","last_page":"5129"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","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"}},"topics":[{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","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"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9962999820709229,"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":0.9951000213623047,"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":0.8227498531341553},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.7931668758392334},{"id":"https://openalex.org/keywords/personalization","display_name":"Personalization","score":0.7792305946350098},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.7615729570388794},{"id":"https://openalex.org/keywords/embedding","display_name":"Embedding","score":0.6779195666313171},{"id":"https://openalex.org/keywords/modular-design","display_name":"Modular design","score":0.6488536596298218},{"id":"https://openalex.org/keywords/adaptation","display_name":"Adaptation (eye)","score":0.5292455554008484},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.48676061630249023},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4347204864025116},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.41968604922294617}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8227498531341553},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.7931668758392334},{"id":"https://openalex.org/C183003079","wikidata":"https://www.wikidata.org/wiki/Q1000371","display_name":"Personalization","level":2,"score":0.7792305946350098},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.7615729570388794},{"id":"https://openalex.org/C41608201","wikidata":"https://www.wikidata.org/wiki/Q980509","display_name":"Embedding","level":2,"score":0.6779195666313171},{"id":"https://openalex.org/C101468663","wikidata":"https://www.wikidata.org/wiki/Q1620158","display_name":"Modular design","level":2,"score":0.6488536596298218},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.5292455554008484},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.48676061630249023},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4347204864025116},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.41968604922294617},{"id":"https://openalex.org/C153294291","wikidata":"https://www.wikidata.org/wiki/Q25261","display_name":"Meteorology","level":1,"score":0.0},{"id":"https://openalex.org/C136764020","wikidata":"https://www.wikidata.org/wiki/Q466","display_name":"World Wide Web","level":1,"score":0.0},{"id":"https://openalex.org/C120665830","wikidata":"https://www.wikidata.org/wiki/Q14620","display_name":"Optics","level":1,"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/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.0},{"id":"https://openalex.org/C77088390","wikidata":"https://www.wikidata.org/wiki/Q8513","display_name":"Database","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.21437/interspeech.2022-10613","is_oa":false,"landing_page_url":"https://doi.org/10.21437/interspeech.2022-10613","pdf_url":null,"source":{"id":"https://openalex.org/S4363604309","display_name":"Interspeech 2022","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":"Interspeech 2022","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":25,"referenced_works":["https://openalex.org/W2060277733","https://openalex.org/W2094147890","https://openalex.org/W2095660652","https://openalex.org/W2101234009","https://openalex.org/W2964303773","https://openalex.org/W2964525587","https://openalex.org/W2970925270","https://openalex.org/W2971840980","https://openalex.org/W2972725946","https://openalex.org/W2972815379","https://openalex.org/W2972970915","https://openalex.org/W3034238904","https://openalex.org/W3097777922","https://openalex.org/W3136987292","https://openalex.org/W3153675281","https://openalex.org/W3161018648","https://openalex.org/W3168867926","https://openalex.org/W3176828726","https://openalex.org/W3194248215","https://openalex.org/W3195577433","https://openalex.org/W3197150384","https://openalex.org/W3201225328","https://openalex.org/W4205991051","https://openalex.org/W4292779060","https://openalex.org/W4322588812"],"related_works":["https://openalex.org/W2108595774","https://openalex.org/W87991986","https://openalex.org/W1576801573","https://openalex.org/W3136652852","https://openalex.org/W1868450377","https://openalex.org/W4221141134","https://openalex.org/W2968378480","https://openalex.org/W4296070216","https://openalex.org/W2963058055","https://openalex.org/W4283396825"],"abstract_inverted_index":{"Model":[0],"fine-tuning":[1],"and":[2,101,110,184],"adaptation":[3,29,84],"have":[4],"become":[5],"a":[6,22,39,42,50,88,153,164,169,178],"common":[7],"approach":[8],"for":[9,12,41,54,72,74,98,113,121,149,155],"model":[10,20,40,99,147],"specialization":[11,36],"downstream":[13],"tasks":[14],"or":[15,21],"domains.":[16],"Fine-tuning":[17,38],"the":[18,25,132,186],"entire":[19],"subset":[23],"of":[24,45,171,188],"parameters":[26],"using":[27,127],"light-weight":[28,83],"has":[30],"shown":[31],"considerable":[32],"success":[33],"across":[34],"different":[35],"tasks.":[37],"large":[43],"number":[44],"domains":[46],"typically":[47],"requires":[48],"starting":[49],"new":[51],"training":[52,100,122,134,190],"job":[53],"every":[55],"domain":[56],"posing":[57],"scaling":[58],"limitations.":[59],"Once":[60],"these":[61],"models":[62],"are":[63],"trained,":[64],"deploying":[65],"them":[66],"also":[67,117,174],"poses":[68],"significant":[69],"scalability":[70,97],"challenges":[71],"inference":[73],"real-time":[75],"applications.":[76],"In":[77],"this":[78],"paper,":[79],"building":[80],"upon":[81],"prior":[82],"techniques,":[85],"we":[86],"propose":[87,118],"modular":[89],"framework":[90,139],"that":[91,106,176],"enables":[92],"us":[93],"to":[94],"substantially":[95],"improve":[96],"inference.":[102,115],"We":[103,116,136,158,173],"introduce":[104],"Submodels":[105,124],"can":[107,181],"be":[108],"quickly":[109],"dynamically":[111],"loaded":[112],"on-the-fly":[114],"multiple":[119],"approaches":[120],"those":[123],"in":[125,131],"parallel":[126],"an":[128,141],"embedding":[129],"space":[130,180],"same":[133],"job.":[135],"test":[137],"our":[138],"on":[140],"extreme":[142],"use-case":[143],"which":[144],"is":[145],"speech":[146],"personalization":[148,189],"atypical":[150],"speech,":[151],"requiring":[152],"Submodel":[154,161],"each":[156],"user.":[157],"obtain":[159],"128x":[160],"throughput":[162],"with":[163],"fixed":[165],"computation":[166],"budget":[167],"without":[168],"loss":[170],"accuracy.":[172],"show":[175],"learning":[177],"speaker-embedding":[179],"scale":[182],"further":[183],"reduce":[185],"amount":[187],"data":[191],"required":[192],"per":[193],"speaker.":[194]},"counts_by_year":[{"year":2025,"cited_by_count":1},{"year":2024,"cited_by_count":1},{"year":2023,"cited_by_count":3}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2022-09-17T00:00:00"}