{"id":"https://openalex.org/W4372267353","doi":"https://doi.org/10.1109/icassp49357.2023.10096405","title":"Mask the Bias: Improving Domain-Adaptive Generalization of CTC-Based ASR with Internal Language Model Estimation","display_name":"Mask the Bias: Improving Domain-Adaptive Generalization of CTC-Based ASR with Internal Language Model Estimation","publication_year":2023,"publication_date":"2023-05-05","ids":{"openalex":"https://openalex.org/W4372267353","doi":"https://doi.org/10.1109/icassp49357.2023.10096405"},"language":"en","primary_location":{"id":"doi:10.1109/icassp49357.2023.10096405","doi":"https://doi.org/10.1109/icassp49357.2023.10096405","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49357.2023.10096405","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 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"},"volume":null,"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2305.03837","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5036502813","display_name":"Nilaksh Das","orcid":"https://orcid.org/0000-0002-5281-5549","observed_orcids":["https://orcid.org/0000-0002-5281-5549"]},"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":"Nilaksh Das","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/A5014130267","display_name":"Monica Sunkara","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":"Monica Sunkara","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/A5050552820","display_name":"Sravan Bodapati","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":"Sravan Bodapati","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/A5004989240","display_name":"Jinglun Cai","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":"Jinglun Cai","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/A5065734653","display_name":"Devang Kulshreshtha","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":"Devang Kulshreshtha","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/A5109604613","display_name":"Jeff Farris","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":"Jeff Farris","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/A5050325468","display_name":"Katrin Kirchhoff","orcid":"https://orcid.org/0000-0002-6645-6030","observed_orcids":["https://orcid.org/0000-0002-6645-6030"]},"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":"Katrin Kirchhoff","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":0.3535,"has_fulltext":true,"cited_by_count":2,"citation_normalized_percentile":{"value":0.50026508,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":89,"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/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"}},{"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"}}],"keywords":[{"id":"https://openalex.org/keywords/domain-adaptation","display_name":"domain adaptation","score":0.746999979019165},{"id":"https://openalex.org/keywords/out-of-domain-generalization","display_name":"out-of-domain generalization","score":0.8379999995231628},{"id":"https://openalex.org/keywords/zero-shot-domain-adaptation","display_name":"zero-shot domain adaptation","score":0.7960000038146973},{"id":"https://openalex.org/keywords/linguistic-bias","display_name":"language bias","score":0.47999998927116394},{"id":"https://openalex.org/keywords/out-of-vocabulary-words","display_name":"out-of-vocabulary words","score":0.6669999957084656},{"id":"https://openalex.org/keywords/end-to-end-speech-recognition","display_name":"end-to-end speech recognition","score":0.6890000104904175}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8408094048500061},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.640920102596283},{"id":"https://openalex.org/C159877910","wikidata":"https://www.wikidata.org/wiki/Q2202883","display_name":"Autoregressive model","level":2,"score":0.5581000447273254},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.5331869125366211},{"id":"https://openalex.org/C137800194","wikidata":"https://www.wikidata.org/wiki/Q11713455","display_name":"Interpolation (computer graphics)","level":3,"score":0.5324184894561768},{"id":"https://openalex.org/C36503486","wikidata":"https://www.wikidata.org/wiki/Q11235244","display_name":"Domain (mathematical analysis)","level":2,"score":0.5245818495750427},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.4967513680458069},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.455728679895401},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.42295241355895996},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3408679962158203},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.09782475233078003},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.08249816298484802},{"id":"https://openalex.org/C104114177","wikidata":"https://www.wikidata.org/wiki/Q79782","display_name":"Motion (physics)","level":2,"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/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.0}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1109/icassp49357.2023.10096405","doi":"https://doi.org/10.1109/icassp49357.2023.10096405","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49357.2023.10096405","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 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2305.03837","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2305.03837","pdf_url":"https://arxiv.org/pdf/2305.03837","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2305.03837","doi":"https://doi.org/10.48550/arxiv.2305.03837","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2305.03837","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2305.03837","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2305.03837","pdf_url":"https://arxiv.org/pdf/2305.03837","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"public-domain","license_id":"https://openalex.org/licenses/public-domain","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4372267353.pdf","grobid_xml":"https://content.openalex.org/works/W4372267353.grobid-xml"},"referenced_works_count":27,"referenced_works":["https://openalex.org/W1494198834","https://openalex.org/W1922655562","https://openalex.org/W2134800885","https://openalex.org/W2766219058","https://openalex.org/W2883416004","https://openalex.org/W2939111082","https://openalex.org/W2962780374","https://openalex.org/W3007037695","https://openalex.org/W3011956499","https://openalex.org/W3097777922","https://openalex.org/W3119308075","https://openalex.org/W3140235797","https://openalex.org/W3146505093","https://openalex.org/W3152221657","https://openalex.org/W3156731511","https://openalex.org/W3161860537","https://openalex.org/W3162249256","https://openalex.org/W3162431424","https://openalex.org/W3166512924","https://openalex.org/W3167895882","https://openalex.org/W3173051858","https://openalex.org/W3205201903","https://openalex.org/W4224918897","https://openalex.org/W4225963373","https://openalex.org/W4226158189","https://openalex.org/W4297841426","https://openalex.org/W4319862261"],"related_works":["https://openalex.org/W4375868856","https://openalex.org/W4213395120","https://openalex.org/W3163560333","https://openalex.org/W4372266510","https://openalex.org/W4408356075","https://openalex.org/W4283031490","https://openalex.org/W4391021763","https://openalex.org/W4308521145","https://openalex.org/W7139928972","https://openalex.org/W4386228543"],"abstract_inverted_index":{"End-to-end":[0],"ASR":[1,109],"models":[2,35],"trained":[3],"on":[4],"large":[5],"amount":[6],"of":[7,17,53,123,175],"data":[8,166],"tend":[9],"to":[10,29,62,79,118,150,157,160,182,202,207],"be":[11],"implicitly":[12],"biased":[13],"towards":[14],"language":[15,22,51,87],"semantics":[16],"the":[18,48,54,59,67,81,115,124,132,143,173,191],"training":[19],"data.":[20],"Internal":[21],"model":[23,55,82],"estimation":[24],"(ILME)":[25],"has":[26],"been":[27],"proposed":[28,144],"mitigate":[30],"this":[31,99],"bias":[32,194],"for":[33,71,107,130],"autoregressive":[34],"such":[36,84],"as":[37,77,89],"attention-based":[38],"encoder-decoder":[39],"and":[40,50,57,86,152],"RNN-T.":[41],"Typically,":[42],"ILME":[43,105,196],"is":[44,75,170],"performed":[45],"by":[46,127,148,155,205],"modularizing":[47],"acoustic":[49,60,85],"components":[52],"architecture,":[56],"eliminating":[58],"input":[61],"perform":[63],"log-linear":[64],"interpolation":[65],"with":[66,179,195],"text-only":[68],"posterior.":[69],"However,":[70],"CTC-based":[72,108],"ASR,":[73],"it":[74],"not":[76],"straightforward":[78],"decouple":[80],"into":[83],"components,":[88],"CTC":[90],"log-posteriors":[91,129],"are":[92],"computed":[93],"in":[94],"a":[95,103,120],"non-autoregressive":[96],"manner.":[97],"In":[98,172],"work,":[100],"we":[101,187],"propose":[102],"novel":[104],"technique":[106],"models.":[110],"Our":[111],"method":[112],"iteratively":[113],"masks":[114],"audio":[116],"timesteps":[117],"estimate":[119],"pseudo":[121],"log-likelihood":[122],"internal":[125],"LM":[126],"accumulating":[128],"only":[131,164],"masked":[133],"timesteps.":[134],"Extensive":[135],"evaluation":[136],"across":[137],"multiple":[138],"out-of-domain":[139],"datasets":[140],"reveals":[141],"that":[142,189],"approach":[145],"improves":[146],"WER":[147,204],"up":[149,156,206],"9.8%":[151],"OOV":[153],"F1-score":[154],"24.6%":[158],"relative":[159],"Shallow":[161,200],"Fusion,":[162],"when":[163],"text":[165],"from":[167],"target":[168,184],"domain":[169,177,185,193],"available.":[171],"case":[174],"zero-shot":[176],"adaptation,":[178],"no":[180],"access":[181],"any":[183],"data,":[186],"demonstrate":[188],"removing":[190],"source":[192],"can":[197],"still":[198],"outperform":[199],"Fusion":[201],"improve":[203],"9.3%":[208],"relative.":[209]},"counts_by_year":[{"year":2025,"cited_by_count":1},{"year":2024,"cited_by_count":1}],"updated_date":"2026-10-30T12:39:25.321706","created_date":"2023-05-07T00:00:00"}