{"id":"https://openalex.org/W4408353781","doi":"https://doi.org/10.1109/icassp49660.2025.10888735","title":"StableQuant: Layer Adaptive Post-Training Quantization for Speech Foundation Models","display_name":"StableQuant: Layer Adaptive Post-Training Quantization for Speech Foundation Models","publication_year":2025,"publication_date":"2025-03-12","ids":{"openalex":"https://openalex.org/W4408353781","doi":"https://doi.org/10.1109/icassp49660.2025.10888735"},"language":"en","primary_location":{"id":"doi:10.1109/icassp49660.2025.10888735","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49660.2025.10888735","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2025 - 2025 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/A5057917723","display_name":"Yeona Hong","orcid":null},"institutions":[{"id":"https://openalex.org/I193775966","display_name":"Yonsei University","ror":"https://ror.org/01wjejq96","country_code":"KR","type":"education","lineage":["https://openalex.org/I193775966"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Yeona Hong","raw_affiliation_strings":["Yonsei University,Department of Electrical and Electronic Engineering,Seoul,Republic of Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Yonsei University,Department of Electrical and Electronic Engineering,Seoul,Republic of Korea","institution_ids":["https://openalex.org/I193775966"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5075891585","display_name":"Hye-Won Han","orcid":"https://orcid.org/0000-0001-6620-8283"},"institutions":[{"id":"https://openalex.org/I193775966","display_name":"Yonsei University","ror":"https://ror.org/01wjejq96","country_code":"KR","type":"education","lineage":["https://openalex.org/I193775966"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Hyewon Han","raw_affiliation_strings":["Yonsei University,Department of Electrical and Electronic Engineering,Seoul,Republic of Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Yonsei University,Department of Electrical and Electronic Engineering,Seoul,Republic of Korea","institution_ids":["https://openalex.org/I193775966"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5017925728","display_name":"Woojin Chung","orcid":"https://orcid.org/0000-0003-2090-4851"},"institutions":[{"id":"https://openalex.org/I193775966","display_name":"Yonsei University","ror":"https://ror.org/01wjejq96","country_code":"KR","type":"education","lineage":["https://openalex.org/I193775966"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Woo-Jin Chung","raw_affiliation_strings":["Yonsei University,Department of Electrical and Electronic Engineering,Seoul,Republic of Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Yonsei University,Department of Electrical and Electronic Engineering,Seoul,Republic of Korea","institution_ids":["https://openalex.org/I193775966"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5056128107","display_name":"Hong-Goo Kang","orcid":"https://orcid.org/0000-0002-6554-0783"},"institutions":[{"id":"https://openalex.org/I193775966","display_name":"Yonsei University","ror":"https://ror.org/01wjejq96","country_code":"KR","type":"education","lineage":["https://openalex.org/I193775966"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Hong-Goo Kang","raw_affiliation_strings":["Yonsei University,Department of Electrical and Electronic Engineering,Seoul,Republic of Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Yonsei University,Department of Electrical and Electronic Engineering,Seoul,Republic of Korea","institution_ids":["https://openalex.org/I193775966"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I193775966"],"apc_list":null,"apc_paid":null,"fwci":1.396,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":{"value":0.82612851,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":94,"max":97},"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":0.9987999796867371,"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":0.9987999796867371,"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":0.9575999975204468,"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":0.9455999732017517,"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/computer-science","display_name":"Computer science","score":0.7178177237510681},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.6246755123138428},{"id":"https://openalex.org/keywords/foundation","display_name":"Foundation (evidence)","score":0.5941483974456787},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"Speech recognition","score":0.5456538796424866},{"id":"https://openalex.org/keywords/layer","display_name":"Layer (electronics)","score":0.4692404568195343},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.4445981979370117},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.39759528636932373},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.13097453117370605},{"id":"https://openalex.org/keywords/materials-science","display_name":"Materials science","score":0.06360119581222534}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7178177237510681},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.6246755123138428},{"id":"https://openalex.org/C2780966255","wikidata":"https://www.wikidata.org/wiki/Q5474306","display_name":"Foundation (evidence)","level":2,"score":0.5941483974456787},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.5456538796424866},{"id":"https://openalex.org/C2779227376","wikidata":"https://www.wikidata.org/wiki/Q6505497","display_name":"Layer (electronics)","level":2,"score":0.4692404568195343},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.4445981979370117},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.39759528636932373},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.13097453117370605},{"id":"https://openalex.org/C192562407","wikidata":"https://www.wikidata.org/wiki/Q228736","display_name":"Materials science","level":0,"score":0.06360119581222534},{"id":"https://openalex.org/C166957645","wikidata":"https://www.wikidata.org/wiki/Q23498","display_name":"Archaeology","level":1,"score":0.0},{"id":"https://openalex.org/C159985019","wikidata":"https://www.wikidata.org/wiki/Q181790","display_name":"Composite material","level":1,"score":0.0},{"id":"https://openalex.org/C153294291","wikidata":"https://www.wikidata.org/wiki/Q25261","display_name":"Meteorology","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},{"id":"https://openalex.org/C95457728","wikidata":"https://www.wikidata.org/wiki/Q309","display_name":"History","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icassp49660.2025.10888735","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49660.2025.10888735","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2025 - 2025 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":24,"referenced_works":["https://openalex.org/W1494198834","https://openalex.org/W2973049979","https://openalex.org/W3034940165","https://openalex.org/W3137147200","https://openalex.org/W3197580070","https://openalex.org/W3203140070","https://openalex.org/W3209059054","https://openalex.org/W3209984917","https://openalex.org/W3215615641","https://openalex.org/W4224821750","https://openalex.org/W4224933800","https://openalex.org/W4281492411","https://openalex.org/W4375869065","https://openalex.org/W4390075359","https://openalex.org/W6739901393","https://openalex.org/W6753069482","https://openalex.org/W6767298317","https://openalex.org/W6771812881","https://openalex.org/W6776320331","https://openalex.org/W6776767859","https://openalex.org/W6780218876","https://openalex.org/W6796815506","https://openalex.org/W6847478871","https://openalex.org/W6862863749"],"related_works":["https://openalex.org/W2381393187","https://openalex.org/W2332779545","https://openalex.org/W230091440","https://openalex.org/W2233261550","https://openalex.org/W2358060160","https://openalex.org/W2035483685","https://openalex.org/W2810751659","https://openalex.org/W258997015","https://openalex.org/W1969764885","https://openalex.org/W2997094352"],"abstract_inverted_index":{"In":[0],"this":[1],"paper,":[2],"we":[3],"propose":[4],"StableQuant,":[5],"a":[6,125],"novel":[7],"adaptive":[8],"post-training":[9],"quantization":[10,63,76],"(PTQ)":[11],"algorithm":[12,93],"for":[13,26,57,78,100],"widely":[14],"used":[15],"speech":[16,103],"foundation":[17],"models":[18,30,123],"(SFMs).":[19],"While":[20],"PTQ":[21,114],"has":[22],"been":[23],"successfully":[24,117],"employed":[25],"compressing":[27],"large":[28],"language":[29],"(LLMs)":[31],"due":[32],"to":[33,36,44,112,124,141],"its":[34],"ability":[35],"bypass":[37],"additional":[38],"fine-tuning,":[39],"directly":[40],"applying":[41],"these":[42],"techniques":[43],"SFMs":[45,52],"may":[46],"not":[47],"yield":[48],"optimal":[49,62],"results,":[50],"as":[51,71],"utilize":[53],"distinct":[54],"network":[55,68],"architecture":[56,69],"feature":[58],"extraction.":[59],"StableQuant":[60,116],"demonstrates":[61],"performance":[64,110,139],"regardless":[65],"of":[66,121],"the":[67,75,84,119,129,134],"type,":[70],"it":[72],"adaptively":[73],"determines":[74],"range":[77],"each":[79],"layer":[80],"by":[81],"analyzing":[82],"both":[83],"scale":[85],"distributions":[86],"and":[87,98,107,127],"overall":[88],"performance.":[89],"We":[90],"evaluate":[91],"our":[92],"on":[94],"two":[95],"SFMs,":[96],"HuBERT":[97],"wav2vec2.0,":[99],"an":[101],"automatic":[102],"recognition":[104],"(ASR)":[105],"task,":[106],"achieve":[108],"superior":[109],"compared":[111],"traditional":[113],"methods.":[115],"reduces":[118],"sizes":[120],"SFM":[122],"quarter":[126],"doubles":[128],"inference":[130],"speed":[131],"while":[132],"limiting":[133],"word":[135],"error":[136],"rate":[137],"(WER)":[138],"drop":[140],"less":[142],"than":[143],"0.3%":[144],"with":[145],"8-bit":[146],"quantization.":[147]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2025-12-28T23:10:05.387466","created_date":"2025-10-10T00:00:00"}