{"id":"https://openalex.org/W3097075707","doi":"https://doi.org/10.21437/interspeech.2020-1363","title":"Weak-Attention Suppression for Transformer Based Speech Recognition","display_name":"Weak-Attention Suppression for Transformer Based Speech Recognition","publication_year":2020,"publication_date":"2020-10-25","ids":{"openalex":"https://openalex.org/W3097075707","doi":"https://doi.org/10.21437/interspeech.2020-1363","mag":"3097075707"},"language":"en","primary_location":{"id":"doi:10.21437/interspeech.2020-1363","is_oa":false,"landing_page_url":"https://doi.org/10.21437/interspeech.2020-1363","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Interspeech 2020","raw_type":"proceedings-article"},"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/A5103247973","display_name":"Yangyang Shi","orcid":"https://orcid.org/0000-0001-5297-4155"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yangyang Shi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100339989","display_name":"Yongqiang Wang","orcid":"https://orcid.org/0000-0002-8600-0276"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yongqiang Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101579713","display_name":"Chunyang Wu","orcid":"https://orcid.org/0000-0002-0269-3555"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chunyang Wu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5047073253","display_name":"Christian Fuegen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Christian Fuegen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101984914","display_name":"Frank Zhang","orcid":"https://orcid.org/0000-0001-9946-6502"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Frank Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103098852","display_name":"Duc Le","orcid":"https://orcid.org/0000-0001-9490-2563"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Duc Le","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5087421101","display_name":"Ching-Feng Yeh","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ching-Feng Yeh","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5041313589","display_name":"Michael L. Seltzer","orcid":"https://orcid.org/0000-0003-3474-2451"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Michael L. Seltzer","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":1.725,"has_fulltext":false,"cited_by_count":19,"citation_normalized_percentile":{"value":0.87932517,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":91,"max":98},"biblio":{"volume":null,"issue":null,"first_page":"4996","last_page":"5000"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.9995999932289124,"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":0.9995999932289124,"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":0.9947999715805054,"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/T10860","display_name":"Speech and Audio Processing","score":0.9930999875068665,"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/transformer","display_name":"Transformer","score":0.8283929824829102},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.720483660697937},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"Speech recognition","score":0.658133864402771},{"id":"https://openalex.org/keywords/word-error-rate","display_name":"Word error rate","score":0.5879911780357361},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.5634685158729553},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.5322948098182678},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3963369131088257},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.3510761857032776},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.10407790541648865}],"concepts":[{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.8283929824829102},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.720483660697937},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.658133864402771},{"id":"https://openalex.org/C40969351","wikidata":"https://www.wikidata.org/wiki/Q3516228","display_name":"Word error rate","level":2,"score":0.5879911780357361},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.5634685158729553},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.5322948098182678},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3963369131088257},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3510761857032776},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.10407790541648865},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.0},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","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/C13280743","wikidata":"https://www.wikidata.org/wiki/Q131089","display_name":"Geodesy","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.21437/interspeech.2020-1363","is_oa":false,"landing_page_url":"https://doi.org/10.21437/interspeech.2020-1363","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Interspeech 2020","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.6600000262260437,"display_name":"Quality Education","id":"https://metadata.un.org/sdg/4"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W2309273277","https://openalex.org/W4288365749","https://openalex.org/W2936497627","https://openalex.org/W3013624417","https://openalex.org/W2061937230","https://openalex.org/W4287826556","https://openalex.org/W1574295218","https://openalex.org/W3098382480","https://openalex.org/W4287598411","https://openalex.org/W3100913109"],"abstract_inverted_index":{"Transformers,":[0],"originally":[1],"proposed":[2,90],"for":[3,102],"natural":[4],"language":[5],"processing":[6],"(NLP)":[7],"tasks,":[8],"have":[9],"recently":[10],"achieved":[11],"great":[12],"success":[13],"in":[14,64,106,147],"automatic":[15],"speech":[16],"recognition":[17],"(ASR).":[18],"However,":[19],"adjacent":[20],"acoustic":[21,127],"units":[22],"(i.e.,":[23],"frames)":[24],"are":[25,33],"highly":[26],"correlated,":[27],"and":[28,47,98,124,129],"long-distance":[29],"dependencies":[30],"between":[31],"them":[32],"weak,":[34],"unlike":[35],"text":[36],"units.":[37],"It":[38,141],"suggests":[39],"that":[40,60,69,116],"ASR":[41,149],"will":[42],"likely":[43,132],"benefit":[44],"from":[45],"sparse":[46],"localized":[48],"attention.":[49],"In":[50],"this":[51],"paper,":[52],"we":[53],"propose":[54],"Weak-Attention":[55],"Suppression":[56],"(WAS),":[57],"a":[58,107],"method":[59,91],"dynamically":[61],"induces":[62],"sparsity":[63],"attention":[65,121],"probabilities.":[66],"We":[67],"demonstrate":[68],"WAS":[70,117],"leads":[71],"to":[72,119,133],"consistent":[73],"Word":[74],"Error":[75],"Rate":[76],"(WER)":[77],"improvement":[78],"over":[79],"strong":[80],"transformer":[81],"baselines.":[82],"On":[83],"the":[84,143],"widely":[85],"used":[86],"LibriSpeech":[87],"benchmark,":[88],"our":[89],"reduced":[92],"WER":[93],"by":[94],"10%$":[95],"on":[96,100],"test-clean":[97],"5%":[99],"test-other":[101],"streamable":[103],"transformers,":[104],"resulting":[105],"new":[108],"state-of-the-art":[109],"among":[110],"streaming":[111],"models.":[112,150],"Further":[113],"analysis":[114],"shows":[115],"learns":[118],"suppress":[120,134],"of":[122,145],"non-critical":[123],"redundant":[125],"continuous":[126],"frames,":[128],"is":[130],"more":[131],"past":[135],"frames":[136],"rather":[137],"than":[138],"future":[139],"ones.":[140],"indicates":[142],"importance":[144],"lookahead":[146],"attention-based":[148]},"counts_by_year":[{"year":2025,"cited_by_count":1},{"year":2024,"cited_by_count":3},{"year":2023,"cited_by_count":3},{"year":2022,"cited_by_count":5},{"year":2021,"cited_by_count":4},{"year":2020,"cited_by_count":3}],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}