{"id":"https://openalex.org/W7148344176","doi":"https://doi.org/10.1109/asru65441.2025.11434657","title":"Group Relative Policy Optimization for Speech Recognition","display_name":"Group Relative Policy Optimization for Speech Recognition","publication_year":2025,"publication_date":"2025-12-06","ids":{"openalex":"https://openalex.org/W7148344176","doi":"https://doi.org/10.1109/asru65441.2025.11434657"},"language":null,"primary_location":{"id":"doi:10.1109/asru65441.2025.11434657","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asru65441.2025.11434657","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)","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/A5088032905","display_name":"Prashanth Gurunath Shivakumar","orcid":"https://orcid.org/0000-0003-1632-3309"},"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":true,"raw_author_name":"Prashanth Gurunath Shivakumar","raw_affiliation_strings":["Amazon Science,Seattle,U.S.A"],"affiliations":[{"raw_affiliation_string":"Amazon Science,Seattle,U.S.A","institution_ids":["https://openalex.org/I1311688040"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5022646825","display_name":"Yile Gu","orcid":"https://orcid.org/0000-0003-3662-5056"},"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":"Yile Gu","raw_affiliation_strings":["Amazon Science,Seattle,U.S.A"],"affiliations":[{"raw_affiliation_string":"Amazon Science,Seattle,U.S.A","institution_ids":["https://openalex.org/I1311688040"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5014054572","display_name":"Ankur Gandhe","orcid":null},"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":"Ankur Gandhe","raw_affiliation_strings":["Amazon Science,Seattle,U.S.A"],"affiliations":[{"raw_affiliation_string":"Amazon Science,Seattle,U.S.A","institution_ids":["https://openalex.org/I1311688040"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5109934165","display_name":"Ivan Bulyko","orcid":null},"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":"Ivan Bulyko","raw_affiliation_strings":["Amazon Science,Seattle,U.S.A"],"affiliations":[{"raw_affiliation_string":"Amazon Science,Seattle,U.S.A","institution_ids":["https://openalex.org/I1311688040"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5088032905"],"corresponding_institution_ids":["https://openalex.org/I1311688040"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.87554716,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"7"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.5569000244140625,"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":0.5569000244140625,"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":0.05730000138282776,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.04610000178217888,"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/leverage","display_name":"Leverage (statistics)","score":0.6977999806404114},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.6485999822616577},{"id":"https://openalex.org/keywords/word-error-rate","display_name":"Word error rate","score":0.532800018787384},{"id":"https://openalex.org/keywords/scalability","display_name":"Scalability","score":0.49140000343322754},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.47119998931884766},{"id":"https://openalex.org/keywords/autoregressive-model","display_name":"Autoregressive model","score":0.4406999945640564},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.351500004529953}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7627999782562256},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.6977999806404114},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.6960999965667725},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.6485999822616577},{"id":"https://openalex.org/C40969351","wikidata":"https://www.wikidata.org/wiki/Q3516228","display_name":"Word error rate","level":2,"score":0.532800018787384},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5015000104904175},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.49140000343322754},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.47119998931884766},{"id":"https://openalex.org/C159877910","wikidata":"https://www.wikidata.org/wiki/Q2202883","display_name":"Autoregressive model","level":2,"score":0.4406999945640564},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4242999851703644},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.351500004529953},{"id":"https://openalex.org/C23224414","wikidata":"https://www.wikidata.org/wiki/Q176769","display_name":"Hidden Markov model","level":2,"score":0.3264999985694885},{"id":"https://openalex.org/C61328038","wikidata":"https://www.wikidata.org/wiki/Q3358061","display_name":"Speech processing","level":2,"score":0.32580000162124634},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.310699999332428},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.31060001254081726},{"id":"https://openalex.org/C36503486","wikidata":"https://www.wikidata.org/wiki/Q11235244","display_name":"Domain (mathematical analysis)","level":2,"score":0.2964000105857849},{"id":"https://openalex.org/C204201278","wikidata":"https://www.wikidata.org/wiki/Q1332614","display_name":"Voice activity detection","level":3,"score":0.2782999873161316},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.27709999680519104},{"id":"https://openalex.org/C150856459","wikidata":"https://www.wikidata.org/wiki/Q8034367","display_name":"Word recognition","level":3,"score":0.26969999074935913},{"id":"https://openalex.org/C90805587","wikidata":"https://www.wikidata.org/wiki/Q10944557","display_name":"Word (group theory)","level":2,"score":0.2630999982357025}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/asru65441.2025.11434657","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asru65441.2025.11434657","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.6251669526100159,"display_name":"Peace, Justice and strong institutions","id":"https://metadata.un.org/sdg/16"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":23,"referenced_works":["https://openalex.org/W2025037086","https://openalex.org/W2125336414","https://openalex.org/W2155004026","https://openalex.org/W2799473636","https://openalex.org/W2951444698","https://openalex.org/W2963796169","https://openalex.org/W2963943581","https://openalex.org/W3095410713","https://openalex.org/W3119308075","https://openalex.org/W3196509775","https://openalex.org/W3210177631","https://openalex.org/W4319862635","https://openalex.org/W4382202533","https://openalex.org/W4389518985","https://openalex.org/W4389524500","https://openalex.org/W4391021666","https://openalex.org/W4392903956","https://openalex.org/W4392904805","https://openalex.org/W4392931626","https://openalex.org/W4412944702","https://openalex.org/W4412944867","https://openalex.org/W4415433258","https://openalex.org/W7133200456"],"related_works":[],"abstract_inverted_index":{"Speech":[0],"Recognition":[1],"has":[2],"seen":[3],"a":[4],"dramatic":[5],"shift":[6,14],"towards":[7],"adopting":[8],"Large":[9],"Language":[10],"Models":[11],"(LLMs).":[12],"This":[13],"is":[15],"partly":[16],"driven":[17],"by":[18,23],"good":[19],"scalability":[20],"properties":[21],"demonstrated":[22],"LLMs,":[24,58],"ability":[25],"to":[26,80,99],"leverage":[27],"large":[28],"amounts":[29],"of":[30,48,74],"labelled,":[31],"unlabelled":[32],"speech":[33,89],"and":[34,42,64,123],"text":[35],"data,":[36],"streaming":[37],"capabilities":[38],"with":[39,44,57,66],"autoregressive":[40],"framework":[41],"multi-tasking":[43],"instruction":[45],"following":[46],"characteristics":[47],"LLMs.":[49],"However,":[50],"simple":[51,94],"next-token":[52],"prediction":[53],"objective,":[54],"typically":[55],"employed":[56],"have":[59],"certain":[60],"limitations":[61],"in":[62,108,116,125],"performance":[63],"challenges":[65],"hallucinations.":[67],"In":[68],"this":[69],"paper,":[70],"we":[71],"propose":[72],"application":[73],"Group":[75],"Relative":[76],"Policy":[77],"Optimization":[78],"(GRPO)":[79],"enable":[81],"reinforcement":[82],"learning":[83],"from":[84],"human":[85],"feedback":[86],"for":[87],"automatic":[88],"recognition":[90],"(ASR).":[91],"We":[92,104],"design":[93],"rule":[95],"based":[96],"reward":[97],"functions":[98],"guide":[100],"the":[101],"policy":[102],"updates.":[103],"demonstrate":[105],"significant":[106],"improvements":[107],"word":[109],"error":[110],"rate":[111],"(upto":[112],"18.4%":[113],"relative),":[114],"reduction":[115],"hallucinations,":[117],"increased":[118],"robustness":[119],"on":[120],"out-of-domain":[121],"datasets":[122],"effectiveness":[124],"domain":[126],"adaptation.":[127]},"counts_by_year":[],"updated_date":"2026-04-03T16:44:17.987007","created_date":"2026-04-03T00:00:00"}