{"id":"https://openalex.org/W7148398150","doi":"https://doi.org/10.1109/asru65441.2025.11434645","title":"Omni-Router: Sharing Routing Decisions in Sparse Mixture-of-Experts for Speech Recognition","display_name":"Omni-Router: Sharing Routing Decisions in Sparse Mixture-of-Experts for Speech Recognition","publication_year":2025,"publication_date":"2025-12-06","ids":{"openalex":"https://openalex.org/W7148398150","doi":"https://doi.org/10.1109/asru65441.2025.11434645"},"language":null,"primary_location":{"id":"doi:10.1109/asru65441.2025.11434645","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asru65441.2025.11434645","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/A5031982816","display_name":"Zijin Gu","orcid":"https://orcid.org/0000-0002-0385-2677"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Zijin Gu","raw_affiliation_strings":["Apple Inc"],"affiliations":[{"raw_affiliation_string":"Apple Inc","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5132816340","display_name":"Tatiana Likhomanenko","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tatiana Likhomanenko","raw_affiliation_strings":["Apple Inc"],"affiliations":[{"raw_affiliation_string":"Apple Inc","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5112445699","display_name":"Navdeep Jaitly","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Navdeep Jaitly","raw_affiliation_strings":["Apple Inc"],"affiliations":[{"raw_affiliation_string":"Apple Inc","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5031982816"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.87562216,"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":"8"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.4846999943256378,"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":0.4846999943256378,"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":0.16979999840259552,"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/T11704","display_name":"Mobile Crowdsensing and Crowdsourcing","score":0.04349999874830246,"subfield":{"id":"https://openalex.org/subfields/1706","display_name":"Computer Science Applications"},"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/router","display_name":"Router","score":0.6261000037193298},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.5630999803543091},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.54339998960495},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.3905999958515167},{"id":"https://openalex.org/keywords/routing-algorithm","display_name":"Routing algorithm","score":0.296099990606308},{"id":"https://openalex.org/keywords/speech-processing","display_name":"Speech processing","score":0.28459998965263367}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7416999936103821},{"id":"https://openalex.org/C2775896111","wikidata":"https://www.wikidata.org/wiki/Q642560","display_name":"Router","level":2,"score":0.6261000037193298},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.5630999803543091},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.54339998960495},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.4699000120162964},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.39410001039505005},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3905999958515167},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3774999976158142},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.3154999911785126},{"id":"https://openalex.org/C2984173633","wikidata":"https://www.wikidata.org/wiki/Q22725","display_name":"Routing algorithm","level":4,"score":0.296099990606308},{"id":"https://openalex.org/C61328038","wikidata":"https://www.wikidata.org/wiki/Q3358061","display_name":"Speech processing","level":2,"score":0.28459998965263367},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.2770000100135803},{"id":"https://openalex.org/C74172769","wikidata":"https://www.wikidata.org/wiki/Q1446839","display_name":"Routing (electronic design automation)","level":2,"score":0.26829999685287476},{"id":"https://openalex.org/C110875604","wikidata":"https://www.wikidata.org/wiki/Q75","display_name":"The Internet","level":2,"score":0.2632000148296356},{"id":"https://openalex.org/C2776604539","wikidata":"https://www.wikidata.org/wiki/Q6423395","display_name":"Knowledge sharing","level":2,"score":0.2632000148296356},{"id":"https://openalex.org/C195358072","wikidata":"https://www.wikidata.org/wiki/Q944584","display_name":"Telephony","level":2,"score":0.26179999113082886},{"id":"https://openalex.org/C204201278","wikidata":"https://www.wikidata.org/wiki/Q1332614","display_name":"Voice activity detection","level":3,"score":0.2578999996185303},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.25220000743865967}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/asru65441.2025.11434645","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asru65441.2025.11434645","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":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":31,"referenced_works":["https://openalex.org/W108866686","https://openalex.org/W1494198834","https://openalex.org/W2024490156","https://openalex.org/W2025653905","https://openalex.org/W2083751884","https://openalex.org/W2150884987","https://openalex.org/W2166637769","https://openalex.org/W2526425061","https://openalex.org/W2799473636","https://openalex.org/W2799800213","https://openalex.org/W2916997151","https://openalex.org/W2936774411","https://openalex.org/W2995181338","https://openalex.org/W3020336359","https://openalex.org/W3097777922","https://openalex.org/W3119308075","https://openalex.org/W3196950043","https://openalex.org/W3197223534","https://openalex.org/W4224299101","https://openalex.org/W4319862635","https://openalex.org/W4372349677","https://openalex.org/W4385822293","https://openalex.org/W4385823078","https://openalex.org/W4392903704","https://openalex.org/W4393160139","https://openalex.org/W4402352982","https://openalex.org/W4402671950","https://openalex.org/W4402671991","https://openalex.org/W4409362497","https://openalex.org/W4415798217","https://openalex.org/W7133198209"],"related_works":[],"abstract_inverted_index":{"Mixture-of-experts":[0],"(MoE)":[1],"models":[2],"have":[3],"expanded":[4],"from":[5],"language":[6],"modeling":[7],"to":[8,101,132],"automatic":[9],"speech":[10],"recognition":[11],"(ASR).":[12],"Traditional":[13],"MoE":[14,72],"methods,":[15],"such":[16],"as":[17],"the":[18,44,47,54,96],"Switch":[19,111],"Transformer,":[20],"route":[21],"experts":[22,57],"independently":[23],"within":[24],"each":[25],"layer.":[26],"Our":[27],"analysis":[28],"reveals":[29],"that":[30,38,95],"routers":[31,48],"in":[32,49,58],"most":[33],"layers":[34,60],"make":[35],"expert":[36,127],"choices":[37,45],"are":[39],"not":[40],"strongly":[41],"correlated":[42],"with":[43],"of":[46],"other":[50],"layers.":[51,73],"To":[52],"increase":[53],"cooperation":[55],"between":[56],"different":[59,71],"and":[61,87,106,110,121,129],"encourage":[62],"greater":[63],"specialization,":[64],"we":[65],"use":[66],"a":[67,83],"shared":[68],"router":[69],"across":[70,89],"We":[74],"call":[75],"this":[76],"model":[77],"Omni-router":[78,97],"Transformer.":[79],"Extensive":[80],"experiments":[81],"on":[82],"large-scale":[84],"pseudolabeled":[85],"dataset":[86],"evaluations":[88],"10":[90],"diverse,":[91],"out-of-domain":[92],"datasets":[93],"demonstrate":[94],"Transformer":[98,112],"is":[99],"able":[100],"achieve":[102],"lower":[103],"training":[104],"loss":[105],"consistently":[107],"outperform":[108],"dense":[109],"models,":[113],"reducing":[114],"average":[115],"word":[116],"error":[117],"rates":[118],"by":[119],"11.2%":[120],"8.2%,":[122],"respectively,":[123],"while":[124],"promoting":[125],"structured":[126],"usage":[128],"improved":[130],"robustness":[131],"diverse":[133],"data.":[134]},"counts_by_year":[],"updated_date":"2026-04-03T16:44:17.987007","created_date":"2026-04-03T00:00:00"}