{"id":"https://openalex.org/W4392908891","doi":"https://doi.org/10.1109/icassp48485.2024.10446132","title":"Leveraging Large Language Models for Exploiting ASR Uncertainty","display_name":"Leveraging Large Language Models for Exploiting ASR Uncertainty","publication_year":2024,"publication_date":"2024-03-18","ids":{"openalex":"https://openalex.org/W4392908891","doi":"https://doi.org/10.1109/icassp48485.2024.10446132"},"language":"en","primary_location":{"id":"doi:10.1109/icassp48485.2024.10446132","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp48485.2024.10446132","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2024 - 2024 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/A5063425263","display_name":"Pranay Dighe","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":true,"raw_author_name":"Pranay Dighe","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5000468857","display_name":"Yi Su","orcid":"https://orcid.org/0000-0001-8794-7376"},"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":"Yi Su","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5070966822","display_name":"Shangshang Zheng","orcid":"https://orcid.org/0000-0002-6291-5578"},"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":"Shangshang Zheng","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5026897397","display_name":"Yunshu Liu","orcid":"https://orcid.org/0000-0002-1415-3601"},"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":"Yunshu Liu","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5083789644","display_name":"Vineet Garg","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":"Vineet Garg","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5069092461","display_name":"Xiaochuan Niu","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":"Xiaochuan Niu","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5089360428","display_name":"Ahmed H. Tewfik","orcid":"https://orcid.org/0000-0002-2384-4391"},"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":"Ahmed Tewfik","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":7,"corresponding_author_ids":["https://openalex.org/A5063425263"],"corresponding_institution_ids":["https://openalex.org/I4210107260"],"apc_list":null,"apc_paid":null,"fwci":3.1008,"has_fulltext":false,"cited_by_count":9,"citation_normalized_percentile":{"value":0.921141,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":96,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"12231","last_page":"12235"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.914900004863739,"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/T10028","display_name":"Topic Modeling","score":0.914900004863739,"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":0.8037568926811218},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.5108137130737305},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.403689444065094},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.39088040590286255}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8037568926811218},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.5108137130737305},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.403689444065094},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.39088040590286255}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icassp48485.2024.10446132","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp48485.2024.10446132","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2024 - 2024 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":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":44,"referenced_works":["https://openalex.org/W2046932483","https://openalex.org/W2132991150","https://openalex.org/W2511962886","https://openalex.org/W2797583228","https://openalex.org/W2936167188","https://openalex.org/W2981022124","https://openalex.org/W3081168214","https://openalex.org/W3097777922","https://openalex.org/W3119913666","https://openalex.org/W3163571828","https://openalex.org/W3170405627","https://openalex.org/W3177813494","https://openalex.org/W3196496149","https://openalex.org/W3198035615","https://openalex.org/W4205385624","https://openalex.org/W4212774754","https://openalex.org/W4221143046","https://openalex.org/W4226278401","https://openalex.org/W4286769130","https://openalex.org/W4297841617","https://openalex.org/W4322718191","https://openalex.org/W4361866031","https://openalex.org/W4377130946","https://openalex.org/W4377372369","https://openalex.org/W4380353763","https://openalex.org/W4383989131","https://openalex.org/W4385567149","https://openalex.org/W4385823351","https://openalex.org/W4389523957","https://openalex.org/W4389524500","https://openalex.org/W4392669753","https://openalex.org/W4392903956","https://openalex.org/W4393178509","https://openalex.org/W6750665317","https://openalex.org/W6778883912","https://openalex.org/W6796581206","https://openalex.org/W6809646742","https://openalex.org/W6810738896","https://openalex.org/W6850625674","https://openalex.org/W6851513886","https://openalex.org/W6852818750","https://openalex.org/W6852874933","https://openalex.org/W6853249747","https://openalex.org/W6854445763"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052","https://openalex.org/W2382290278","https://openalex.org/W3204019825"],"abstract_inverted_index":{"While":[0],"large":[1],"language":[2,10,19],"models":[3],"excel":[4],"in":[5],"a":[6,63,78,94,190,199],"variety":[7],"of":[8,62,92,104,145,149,162,172],"natural":[9],"processing":[11],"(NLP)":[12],"tasks,":[13,22],"to":[14,86,109,158,165,186,223],"perform":[15],"well":[16,196],"on":[17,27,47,54,67,176,189,198],"spoken":[18,69,89],"understanding":[20],"(SLU)":[21],"they":[23],"must":[24],"either":[25],"rely":[26],"off-the-shelf":[28],"automatic":[29],"speech":[30,42,192],"recognition":[31],"(ASR)":[32],"systems":[33,204],"for":[34,219,229],"transcription,":[35],"or":[36,100],"be":[37,127,187],"equipped":[38],"with":[39,141],"an":[40,142,220],"in-built":[41],"modality.":[43],"This":[44],"work":[45],"focuses":[46],"the":[48,60,68,83,88,119,139,151,160,166,170,177,217],"former":[49],"scenario,":[50],"where":[51,77,203],"LLM\u2019s":[52,84],"accuracy":[53,61,96],"SLU":[55],"tasks":[56],"is":[57],"constrained":[58],"by":[59,97,129,169],"fixed":[64],"ASR":[65,121,146,213,225],"system":[66],"input.":[70],"Specifically,":[71],"we":[72,107,113,136],"tackle":[73],"speech-intent":[74],"classification":[75],"task,":[76,202],"high":[79,95],"word-error-rate":[80],"can":[81,114,125],"limit":[82],"ability":[85],"understand":[87],"intent.":[90],"Instead":[91],"chasing":[93],"designing":[98],"complex":[99],"specialized":[101],"architectures":[102],"regardless":[103],"deployment":[105],"costs,":[106],"seek":[108],"answer":[110],"how":[111],"far":[112],"go":[115],"without":[116],"substantially":[117],"changing":[118],"underlying":[120],"and":[122],"LLM,":[123],"which":[124],"potentially":[126],"shared":[128],"multiple":[130],"unrelated":[131],"tasks.":[132,179],"To":[133],"this":[134],"end,":[135],"propose":[137],"prompting":[138],"LLM":[140],"n-best":[143,163,183,206],"list":[144,207],"hypotheses":[147],"instead":[148],"only":[150],"error-prone":[152],"1-best":[153,212],"hypothesis.":[154],"We":[155],"explore":[156],"prompt-engineering":[157],"explain":[159],"concept":[161],"lists":[164,184],"LLM;":[167],"followed":[168],"finetuning":[171],"Low-Rank":[173],"Adapters":[174],"[1]":[175],"downstream":[178],"Our":[180],"approach":[181],"using":[182,205,211],"proves":[185],"effective":[188],"device-directed":[191],"detection":[193],"task":[194],"as":[195,197],"keyword":[200],"spotting":[201],"prompts":[208],"outperform":[209],"those":[210],"hypothesis;":[214],"thus":[215],"paving":[216],"way":[218],"efficient":[221],"method":[222],"exploit":[224],"uncertainty":[226],"via":[227],"LLMs":[228],"speech-based":[230],"applications.":[231]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":5},{"year":2024,"cited_by_count":3}],"updated_date":"2026-04-03T22:45:19.894376","created_date":"2025-10-10T00:00:00"}