{"id":"https://openalex.org/W4375869390","doi":"https://doi.org/10.1109/icassp49357.2023.10094588","title":"Variable Attention Masking for Configurable Transformer Transducer Speech Recognition","display_name":"Variable Attention Masking for Configurable Transformer Transducer Speech Recognition","publication_year":2023,"publication_date":"2023-05-05","ids":{"openalex":"https://openalex.org/W4375869390","doi":"https://doi.org/10.1109/icassp49357.2023.10094588"},"language":"en","primary_location":{"id":"doi:10.1109/icassp49357.2023.10094588","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49357.2023.10094588","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/A5084169592","display_name":"Pawe\u0142 \u015awi\u0119toja\u0144ski","orcid":"https://orcid.org/0000-0001-5896-4505"},"institutions":[{"id":"https://openalex.org/I4210107260","display_name":"Apple (United Kingdom)","ror":"https://ror.org/01vpeym60","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210107260"]}],"countries":["GB"],"is_corresponding":true,"raw_author_name":"Pawel Swietojanski","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5083904404","display_name":"Stefan Braun","orcid":"https://orcid.org/0000-0002-7145-1103"},"institutions":[{"id":"https://openalex.org/I4210107260","display_name":"Apple (United Kingdom)","ror":"https://ror.org/01vpeym60","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210107260"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Stefan Braun","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5112207449","display_name":"Do\u011fan Can","orcid":null},"institutions":[{"id":"https://openalex.org/I4210107260","display_name":"Apple (United Kingdom)","ror":"https://ror.org/01vpeym60","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210107260"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Dogan Can","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5067994666","display_name":"Thiago Fraga da Silva","orcid":null},"institutions":[{"id":"https://openalex.org/I4210107260","display_name":"Apple (United Kingdom)","ror":"https://ror.org/01vpeym60","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210107260"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Thiago Fraga Da Silva","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5016623654","display_name":"Arnab Ghoshal","orcid":"https://orcid.org/0000-0003-1272-2599"},"institutions":[{"id":"https://openalex.org/I4210107260","display_name":"Apple (United Kingdom)","ror":"https://ror.org/01vpeym60","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210107260"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Arnab Ghoshal","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5087554069","display_name":"Takaaki Hori","orcid":"https://orcid.org/0000-0003-4560-8039"},"institutions":[{"id":"https://openalex.org/I4210107260","display_name":"Apple (United Kingdom)","ror":"https://ror.org/01vpeym60","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210107260"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Takaaki Hori","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5033764781","display_name":"Roger Hsiao","orcid":"https://orcid.org/0000-0001-8638-4508"},"institutions":[{"id":"https://openalex.org/I4210107260","display_name":"Apple (United Kingdom)","ror":"https://ror.org/01vpeym60","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210107260"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Roger Hsiao","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5033446148","display_name":"H. Benjamin Mason","orcid":"https://orcid.org/0000-0003-4279-2854"},"institutions":[{"id":"https://openalex.org/I4210107260","display_name":"Apple (United Kingdom)","ror":"https://ror.org/01vpeym60","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210107260"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Henry Mason","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030109648","display_name":"Erik McDermott","orcid":null},"institutions":[{"id":"https://openalex.org/I4210107260","display_name":"Apple (United Kingdom)","ror":"https://ror.org/01vpeym60","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210107260"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Erik McDermott","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050533390","display_name":"Honza Silovsky","orcid":null},"institutions":[{"id":"https://openalex.org/I4210107260","display_name":"Apple (United Kingdom)","ror":"https://ror.org/01vpeym60","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210107260"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Honza Silovsky","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5079853614","display_name":"Ruchir Travadi","orcid":"https://orcid.org/0000-0002-1272-5282"},"institutions":[{"id":"https://openalex.org/I4210107260","display_name":"Apple (United Kingdom)","ror":"https://ror.org/01vpeym60","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210107260"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Ruchir Travadi","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5110376025","display_name":"Xiaodan Zhuang","orcid":null},"institutions":[{"id":"https://openalex.org/I4210107260","display_name":"Apple (United Kingdom)","ror":"https://ror.org/01vpeym60","country_code":"GB","type":"company","lineage":["https://openalex.org/I4210107260"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Xiaodan Zhuang","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":12,"corresponding_author_ids":["https://openalex.org/A5084169592"],"corresponding_institution_ids":["https://openalex.org/I4210107260"],"apc_list":null,"apc_paid":null,"fwci":1.3986,"has_fulltext":false,"cited_by_count":8,"citation_normalized_percentile":{"value":0.8450487,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":96,"max":98},"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":0.9998999834060669,"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":0.9998999834060669,"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":0.9998000264167786,"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/T11309","display_name":"Music and Audio Processing","score":0.9976000189781189,"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/masking","display_name":"Masking (illustration)","score":0.8199375867843628},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.8012195825576782},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.6397395730018616},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.623314619064331},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"Speech recognition","score":0.5651257634162903},{"id":"https://openalex.org/keywords/auditory-masking","display_name":"Auditory masking","score":0.4661606252193451},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.350680410861969},{"id":"https://openalex.org/keywords/telecommunications","display_name":"Telecommunications","score":0.09386217594146729}],"concepts":[{"id":"https://openalex.org/C2777402240","wikidata":"https://www.wikidata.org/wiki/Q6783436","display_name":"Masking (illustration)","level":2,"score":0.8199375867843628},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8012195825576782},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.6397395730018616},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.623314619064331},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.5651257634162903},{"id":"https://openalex.org/C77801330","wikidata":"https://www.wikidata.org/wiki/Q1363413","display_name":"Auditory masking","level":3,"score":0.4661606252193451},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.350680410861969},{"id":"https://openalex.org/C76155785","wikidata":"https://www.wikidata.org/wiki/Q418","display_name":"Telecommunications","level":1,"score":0.09386217594146729},{"id":"https://openalex.org/C142362112","wikidata":"https://www.wikidata.org/wiki/Q735","display_name":"Art","level":0,"score":0.0},{"id":"https://openalex.org/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.0},{"id":"https://openalex.org/C153349607","wikidata":"https://www.wikidata.org/wiki/Q36649","display_name":"Visual arts","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}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icassp49357.2023.10094588","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49357.2023.10094588","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":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":42,"referenced_works":["https://openalex.org/W1828163288","https://openalex.org/W1836465849","https://openalex.org/W2291513470","https://openalex.org/W2964110616","https://openalex.org/W3007433671","https://openalex.org/W3015686596","https://openalex.org/W3015974384","https://openalex.org/W3016010032","https://openalex.org/W3092122846","https://openalex.org/W3096123040","https://openalex.org/W3096686110","https://openalex.org/W3097030750","https://openalex.org/W3097777922","https://openalex.org/W3112702554","https://openalex.org/W3125815078","https://openalex.org/W3160628828","https://openalex.org/W3161873870","https://openalex.org/W3162665866","https://openalex.org/W3163203022","https://openalex.org/W3163907627","https://openalex.org/W3189164341","https://openalex.org/W3197478142","https://openalex.org/W3197654132","https://openalex.org/W3197956343","https://openalex.org/W3198442913","https://openalex.org/W3211278025","https://openalex.org/W4287124724","https://openalex.org/W4297841343","https://openalex.org/W4385245566","https://openalex.org/W6638667902","https://openalex.org/W6638749077","https://openalex.org/W6739901393","https://openalex.org/W6757998062","https://openalex.org/W6773475747","https://openalex.org/W6775607357","https://openalex.org/W6777191834","https://openalex.org/W6780226713","https://openalex.org/W6784400248","https://openalex.org/W6790121257","https://openalex.org/W6797264745","https://openalex.org/W6804209468","https://openalex.org/W7064788384"],"related_works":["https://openalex.org/W2140659649","https://openalex.org/W2010677552","https://openalex.org/W2218947444","https://openalex.org/W2315293800","https://openalex.org/W2411742137","https://openalex.org/W1602745308","https://openalex.org/W2109316093","https://openalex.org/W2990940109","https://openalex.org/W2055429703","https://openalex.org/W1988918813"],"abstract_inverted_index":{"This":[0],"work":[1,92],"studies":[2],"the":[3,35,48,69,75,147,155],"use":[4,70],"of":[5,29,61,71],"attention":[6,37,49,76],"masking":[7,123,145],"in":[8,59,93,154],"transformer":[9],"transducer":[10],"based":[11],"speech":[12],"recognition":[13,62,113],"for":[14,20,51],"building":[15],"a":[16,26,81,100,125],"single":[17,101],"configurable":[18,102],"model":[19,103],"different":[21,94],"deployment":[22],"scenarios.":[23],"We":[24,66,140],"present":[25],"comprehensive":[27],"set":[28],"experiments":[30],"comparing":[31],"fixed":[32,133],"masking,":[33,46,73,134],"where":[34,47,74],"same":[36],"mask":[38,50],"is":[39,54],"applied":[40],"at":[41,84],"every":[42],"frame,":[43],"with":[44,136],"chunked":[45,122],"each":[52],"frame":[53],"determined":[55],"by":[56,149],"chunk":[57],"boundaries,":[58],"terms":[60],"accuracy":[63,127,148],"and":[64,114,137],"latency.":[65],"then":[67],"explore":[68],"variable":[72,144],"masks":[77],"are":[78],"sampled":[79],"from":[80],"target":[82],"distribution":[83],"training":[85],"time,":[86],"to":[87,107,132,151],"build":[88],"models":[89],"that":[90,121,143],"can":[91,104],"configurations.":[95],"Finally,":[96],"we":[97],"investigate":[98],"how":[99],"be":[105],"used":[106],"perform":[108],"both":[109,135],"first":[110],"pass":[111,116],"streaming":[112],"second":[115],"acoustic":[117,156],"rescoring.":[118],"Experiments":[119],"show":[120,142],"achieves":[124],"better":[126],"vs":[128],"latency":[129],"trade-off":[130],"compared":[131],"without":[138],"FastEmit.":[139],"also":[141],"improves":[146],"up":[150],"8%":[152],"relative":[153],"re-scoring":[157],"scenario.":[158]},"counts_by_year":[{"year":2025,"cited_by_count":5},{"year":2024,"cited_by_count":3}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}