{"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","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49357.2023.10096405","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/A5036502813","display_name":"Nilaksh Das","orcid":"https://orcid.org/0000-0002-5281-5549"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Nilaksh Das","raw_affiliation_strings":["AWS AI Labs,USA","AWS AI Labs, USA"],"affiliations":[{"raw_affiliation_string":"AWS AI Labs,USA","institution_ids":[]},{"raw_affiliation_string":"AWS AI Labs, USA","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5014130267","display_name":"Monica Sunkara","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Monica Sunkara","raw_affiliation_strings":["AWS AI Labs,USA","AWS AI Labs, USA"],"affiliations":[{"raw_affiliation_string":"AWS AI Labs,USA","institution_ids":[]},{"raw_affiliation_string":"AWS AI Labs, USA","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050552820","display_name":"Sravan Bodapati","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sravan Bodapati","raw_affiliation_strings":["AWS AI Labs,USA","AWS AI Labs, USA"],"affiliations":[{"raw_affiliation_string":"AWS AI Labs,USA","institution_ids":[]},{"raw_affiliation_string":"AWS AI Labs, USA","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5004989240","display_name":"Jinglun Cai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jinglun Cai","raw_affiliation_strings":["AWS AI Labs,USA","AWS AI Labs, USA"],"affiliations":[{"raw_affiliation_string":"AWS AI Labs,USA","institution_ids":[]},{"raw_affiliation_string":"AWS AI Labs, USA","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5065734653","display_name":"Devang Kulshreshtha","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Devang Kulshreshtha","raw_affiliation_strings":["AWS AI Labs,USA","AWS AI Labs, USA"],"affiliations":[{"raw_affiliation_string":"AWS AI Labs,USA","institution_ids":[]},{"raw_affiliation_string":"AWS AI Labs, USA","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5109604613","display_name":"Jeff Farris","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jeff Farris","raw_affiliation_strings":["AWS AI Labs,USA","AWS AI Labs, USA"],"affiliations":[{"raw_affiliation_string":"AWS AI Labs,USA","institution_ids":[]},{"raw_affiliation_string":"AWS AI Labs, USA","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5050325468","display_name":"Katrin Kirchhoff","orcid":"https://orcid.org/0000-0002-6645-6030"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Katrin Kirchhoff","raw_affiliation_strings":["AWS AI Labs,USA","AWS AI Labs, USA"],"affiliations":[{"raw_affiliation_string":"AWS AI Labs,USA","institution_ids":[]},{"raw_affiliation_string":"AWS AI Labs, USA","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":7,"corresponding_author_ids":["https://openalex.org/A5036502813"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.3497,"has_fulltext":false,"cited_by_count":2,"citation_normalized_percentile":{"value":0.63057892,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":90,"max":95},"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":0.9998000264167786,"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":0.9998000264167786,"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":0.9958999752998352,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9957000017166138,"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":0.8408094048500061},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.640920102596283},{"id":"https://openalex.org/keywords/autoregressive-model","display_name":"Autoregressive model","score":0.5581000447273254},{"id":"https://openalex.org/keywords/generalization","display_name":"Generalization","score":0.5331869125366211},{"id":"https://openalex.org/keywords/interpolation","display_name":"Interpolation (computer graphics)","score":0.5324184894561768},{"id":"https://openalex.org/keywords/domain","display_name":"Domain (mathematical analysis)","score":0.5245818495750427},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"Speech recognition","score":0.4967513680458069},{"id":"https://openalex.org/keywords/encoder","display_name":"Encoder","score":0.455728679895401},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.42295241355895996},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.3408679962158203},{"id":"https://openalex.org/keywords/statistics","display_name":"Statistics","score":0.09782475233078003},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.08249816298484802}],"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":1,"locations":[{"id":"doi:10.1109/icassp49357.2023.10096405","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49357.2023.10096405","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":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality Education","score":0.7799999713897705}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":32,"referenced_works":["https://openalex.org/W1494198834","https://openalex.org/W1922655562","https://openalex.org/W2134800885","https://openalex.org/W2766219058","https://openalex.org/W2939111082","https://openalex.org/W2962780374","https://openalex.org/W2963362078","https://openalex.org/W3008037978","https://openalex.org/W3016234571","https://openalex.org/W3097777922","https://openalex.org/W3119308075","https://openalex.org/W3140235797","https://openalex.org/W3146505093","https://openalex.org/W3152221657","https://openalex.org/W3161860537","https://openalex.org/W3162249256","https://openalex.org/W3162431424","https://openalex.org/W3167895882","https://openalex.org/W3173051858","https://openalex.org/W3186813771","https://openalex.org/W3197688480","https://openalex.org/W3198439131","https://openalex.org/W3205201903","https://openalex.org/W4224918897","https://openalex.org/W4226158189","https://openalex.org/W4296068796","https://openalex.org/W4297841426","https://openalex.org/W4319862261","https://openalex.org/W6629717138","https://openalex.org/W6640090968","https://openalex.org/W6679855610","https://openalex.org/W6796656850"],"related_works":["https://openalex.org/W2171218219","https://openalex.org/W1972271943","https://openalex.org/W2150410159","https://openalex.org/W3162204513","https://openalex.org/W3150905897","https://openalex.org/W4327525404","https://openalex.org/W4287185323","https://openalex.org/W1520183331","https://openalex.org/W2099889858","https://openalex.org/W2168175994"],"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":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}