{"id":"https://openalex.org/W4408354530","doi":"https://doi.org/10.1109/icassp49660.2025.10887809","title":"Hyper-adapter for Parameter-Efficient Multilingual ASR Adaptation","display_name":"Hyper-adapter for Parameter-Efficient Multilingual ASR Adaptation","publication_year":2025,"publication_date":"2025-03-12","ids":{"openalex":"https://openalex.org/W4408354530","doi":"https://doi.org/10.1109/icassp49660.2025.10887809"},"language":"en","primary_location":{"id":"doi:10.1109/icassp49660.2025.10887809","doi":"https://doi.org/10.1109/icassp49660.2025.10887809","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49660.2025.10887809","pdf_url":null,"source":{"id":"https://openalex.org/S4363607879","display_name":"IEEE International Conference on Acoustics Speech and Signal Processing","issn_l":"1520-6149","issn":["1520-6149","2379-190X"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":["jufo-2","norway-1"],"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":"ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"},"volume":null,"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/A5055234106","display_name":"Zejiang Hou","orcid":"https://orcid.org/0000-0002-6836-8288","observed_orcids":["https://orcid.org/0000-0002-6836-8288"]},"institutions":[{"id":"https://openalex.org/I1311688040","display_name":"Amazon (United States)","ror":"https://ror.org/04mv4n011","country_code":"US","type":"company","lineage":["https://openalex.org/I1311688040"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Zejiang Hou","raw_affiliation_strings":["AWS AI Labs,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"AWS AI Labs,USA","institution_ids":["https://openalex.org/I1311688040"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5000143881","display_name":"Daniel Garcia-Romero","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1311688040","display_name":"Amazon (United States)","ror":"https://ror.org/04mv4n011","country_code":"US","type":"company","lineage":["https://openalex.org/I1311688040"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Daniel Garcia-Romero","raw_affiliation_strings":["AWS AI Labs,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"AWS AI Labs,USA","institution_ids":["https://openalex.org/I1311688040"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5106859201","display_name":"Kyu Jeong Han","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1311688040","display_name":"Amazon (United States)","ror":"https://ror.org/04mv4n011","country_code":"US","type":"company","lineage":["https://openalex.org/I1311688040"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Kyu J. Han","raw_affiliation_strings":["AWS AI Labs,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"AWS AI Labs,USA","institution_ids":["https://openalex.org/I1311688040"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I1311688040"],"apc_list":null,"apc_paid":null,"fwci":0.9884,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":{"value":0.67576866,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":90,"max":94},"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":0.9994999766349792,"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":0.9994999766349792,"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/T10860","display_name":"Speech and Audio Processing","score":9.999999747378752e-05,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":9.999999747378752e-05,"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/multilingual-speech-recognition","display_name":"multilingual speech recognition","score":0.8410000205039978},{"id":"https://openalex.org/keywords/parameter-efficient-fine-tuning","display_name":"parameter-efficient fine-tuning","score":0.9509999752044678},{"id":"https://openalex.org/keywords/hypernetworks","display_name":"hypernetworks","score":0.8790000081062317},{"id":"https://openalex.org/keywords/residual-adapters","display_name":"residual adapters","score":0.777999997138977},{"id":"https://openalex.org/keywords/weight-sharing","display_name":"parameter sharing","score":0.6759999990463257},{"id":"https://openalex.org/keywords/orthogonal-regularization","display_name":"orthogonal regularization","score":0.9959999918937683}],"concepts":[{"id":"https://openalex.org/C177284502","wikidata":"https://www.wikidata.org/wiki/Q1005390","display_name":"Adapter (computing)","level":2,"score":0.9327966570854187},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.7261797189712524},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.710746169090271},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.3787202537059784},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.17178261280059814},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.07745912671089172},{"id":"https://openalex.org/C169760540","wikidata":"https://www.wikidata.org/wiki/Q207011","display_name":"Neuroscience","level":1,"score":0.04205799102783203}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icassp49660.2025.10887809","doi":"https://doi.org/10.1109/icassp49660.2025.10887809","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49660.2025.10887809","pdf_url":null,"source":{"id":"https://openalex.org/S4363607879","display_name":"IEEE International Conference on Acoustics Speech and Signal Processing","issn_l":"1520-6149","issn":["1520-6149","2379-190X"],"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":["jufo-2","norway-1"],"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":"ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"sustainable_development_goals_aurora":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":25,"referenced_works":["https://openalex.org/W2739883972","https://openalex.org/W2885185669","https://openalex.org/W2913946806","https://openalex.org/W2970925270","https://openalex.org/W2995929068","https://openalex.org/W3023528699","https://openalex.org/W3095410713","https://openalex.org/W3110524561","https://openalex.org/W3119308075","https://openalex.org/W3161686170","https://openalex.org/W3162249256","https://openalex.org/W3165404421","https://openalex.org/W3167473228","https://openalex.org/W3172698324","https://openalex.org/W3201225328","https://openalex.org/W4221161761","https://openalex.org/W4225274946","https://openalex.org/W4241414120","https://openalex.org/W4319862635","https://openalex.org/W4380993062","https://openalex.org/W4392902792","https://openalex.org/W4402112144","https://openalex.org/W6850218400","https://openalex.org/W6852909395","https://openalex.org/W7064683377"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2133028525","https://openalex.org/W4229060448","https://openalex.org/W4306381730","https://openalex.org/W2981692913","https://openalex.org/W3044188621","https://openalex.org/W3184035966","https://openalex.org/W2485605994"],"abstract_inverted_index":{"This":[0],"work":[1],"proposes":[2],"a":[3,66,74,96],"new":[4],"parameter-efficient":[5],"adaptation":[6,18,140],"approach":[7,125],"for":[8,25,33,80],"multilingual":[9,16,138,145],"ASR":[10,17,139,146],"based":[11],"on":[12,142],"the":[13,27,42,57,78,81,88,111,120],"hyper-network.":[14],"Existing":[15],"methods":[19],"apply":[20],"either":[21],"one":[22],"residual":[23,38],"adapter":[24,39],"all":[26],"languages,":[28,100],"or":[29,128],"language":[30,54,58,116],"dependent":[31,59],"adapters":[32,60,82],"each":[34],"individual":[35],"language.":[36],"The":[37,123],"cannot":[40],"compete":[41],"full":[43],"finetuning":[44],"in":[45],"terms":[46],"of":[47,99],"WER,":[48],"because":[49],"it":[50],"is":[51],"agnostic":[52],"to":[53,76,109,136],"information.":[55],"Whereas":[56],"introduce":[61],"high":[62],"parameter":[63,67,90,133],"overhead":[64],"without":[65],"sharing":[68,91],"strategy.":[69],"In":[70],"contrast,":[71],"we":[72,101],"leverage":[73],"hyper-network":[75,112],"generate":[77],"weights":[79],"across":[83],"different":[84],"languages.":[85],"To":[86],"achieve":[87],"best":[89],"strategy":[92],"that":[93],"scales":[94],"with":[95],"large":[97],"number":[98],"propose":[102],"multi-level":[103],"conditioning":[104],"vector":[105],"fusion,":[106],"orthogonal":[107],"regularization":[108],"improve":[110],"output":[113],"diversity,":[114],"and":[115,131],"loss":[117],"weighting":[118],"during":[119],"model":[121],"training.":[122],"proposed":[124],"demonstrates":[126],"comparable":[127],"better":[129,132],"WER":[130],"efficiency":[134],"compared":[135],"previous":[137],"approaches":[141],"commonly":[143],"used":[144],"benchmarks.":[147]},"counts_by_year":[{"year":2025,"cited_by_count":1}],"updated_date":"2026-10-16T09:59:22.235173","created_date":"2025-10-10T00:00:00"}