{"id":"https://openalex.org/W3198643121","doi":"https://doi.org/10.21437/interspeech.2021-1367","title":"FSR: Accelerating the Inference Process of Transducer-Based Models by Applying Fast-Skip Regularization","display_name":"FSR: Accelerating the Inference Process of Transducer-Based Models by Applying Fast-Skip Regularization","publication_year":2021,"publication_date":"2021-08-27","ids":{"openalex":"https://openalex.org/W3198643121","doi":"https://doi.org/10.21437/interspeech.2021-1367","mag":"3198643121"},"language":"en","primary_location":{"id":"doi:10.21437/interspeech.2021-1367","is_oa":false,"landing_page_url":"https://doi.org/10.21437/interspeech.2021-1367","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Interspeech 2021","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/A5101817056","display_name":"Zhengkun Tian","orcid":"https://orcid.org/0000-0002-0469-3049"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhengkun Tian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5078525423","display_name":"Jiangyan Yi","orcid":"https://orcid.org/0000-0003-2422-4618"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jiangyan Yi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100624455","display_name":"Ye Bai","orcid":"https://orcid.org/0000-0001-5533-6909"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ye Bai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5112613657","display_name":"Jianhua Tao","orcid":"https://orcid.org/0000-0002-9344-6428"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jianhua Tao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100328838","display_name":"Shuai Zhang","orcid":"https://orcid.org/0000-0001-8502-2927"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Shuai Zhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5111856667","display_name":"Zhengqi Wen","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhengqi Wen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":6,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":1.2594,"has_fulltext":false,"cited_by_count":10,"citation_normalized_percentile":{"value":0.83946185,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":94,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"4034","last_page":"4038"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.9994000196456909,"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":0.9994000196456909,"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":0.9965000152587891,"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":0.9896000027656555,"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/regularization","display_name":"Regularization (linguistics)","score":0.7653577327728271},{"id":"https://openalex.org/keywords/transducer","display_name":"Transducer","score":0.7137237787246704},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6767548322677612},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6546752452850342},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.543155312538147},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.4002474546432495},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.31623148918151855},{"id":"https://openalex.org/keywords/acoustics","display_name":"Acoustics","score":0.15632525086402893},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.08782365918159485},{"id":"https://openalex.org/keywords/physics","display_name":"Physics","score":0.08360141515731812}],"concepts":[{"id":"https://openalex.org/C2776135515","wikidata":"https://www.wikidata.org/wiki/Q17143721","display_name":"Regularization (linguistics)","level":2,"score":0.7653577327728271},{"id":"https://openalex.org/C56318395","wikidata":"https://www.wikidata.org/wiki/Q215928","display_name":"Transducer","level":2,"score":0.7137237787246704},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6767548322677612},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6546752452850342},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.543155312538147},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4002474546432495},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.31623148918151855},{"id":"https://openalex.org/C24890656","wikidata":"https://www.wikidata.org/wiki/Q82811","display_name":"Acoustics","level":1,"score":0.15632525086402893},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.08782365918159485},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.08360141515731812}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.21437/interspeech.2021-1367","is_oa":false,"landing_page_url":"https://doi.org/10.21437/interspeech.2021-1367","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Interspeech 2021","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.5199999809265137,"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":0,"referenced_works":[],"related_works":["https://openalex.org/W2072454424","https://openalex.org/W4384820447","https://openalex.org/W2185942010","https://openalex.org/W2051487156","https://openalex.org/W2260725127","https://openalex.org/W2073681303","https://openalex.org/W2117438306","https://openalex.org/W1992056405","https://openalex.org/W767846903","https://openalex.org/W4200299863"],"abstract_inverted_index":{"Transducer-based":[0],"models,":[1],"such":[2],"as":[3],"RNN-Transducer":[4],"and":[5,28,58,131,140],"transformer-transducer,":[6],"have":[7],"achieved":[8],"great":[9],"success":[10],"in":[11,41,72,114],"speech":[12],"recognition.":[13],"A":[14],"typical":[15],"transducer":[16,95,107,129,168],"model":[17,108,169],"decodes":[18],"the":[19,24,42,62,67,73,89,104,106,111,128,138,142,161,167,171,176],"output":[20,75],"sequence":[21],"conditioned":[22],"on":[23,150],"current":[25],"acoustic":[26],"state":[27],"previously":[29],"predicted":[30,92,98],"tokens":[31,40,69,113],"step":[32],"by":[33,93,99,116],"step.":[34],"Statistically,":[35],"The":[36,157],"number":[37],"of":[38,49,56,127],"blank":[39,63,90,112,172],"prediction":[43],"results":[44,158],"accounts":[45],"for":[46],"nearly":[47,184],"90\\%":[48],"all":[50],"tokens.":[51],"It":[52],"takes":[53],"a":[54,80,94,100,117,151,189],"lot":[55],"computation":[57,139],"time":[59],"to":[60,87],"predict":[61,110],"tokens,":[64],"but":[65],"only":[66,188],"non-blank":[68],"will":[70,136],"appear":[71],"final":[74],"sequence.":[76],"Therefore,":[77],"we":[78],"propose":[79],"method":[81],"named":[82],"fast-skip":[83,162,179],"regularization,":[84],"which":[85,135],"tries":[86],"align":[88],"position":[91,173],"with":[96,178,187],"that":[97,160],"CTC":[101,119],"model.":[102],"During":[103],"inference,":[105],"can":[109,164,180],"advance":[115],"simple":[118],"project":[120],"layer":[121],"without":[122],"many":[123],"complicated":[124],"forward":[125],"calculations":[126],"decoder":[130],"then":[132],"skip":[133],"them,":[134],"reduce":[137],"improve":[141],"inference":[143,177],"speed":[144],"greatly.":[145],"All":[146],"experiments":[147],"are":[148],"conducted":[149],"public":[152],"Chinese":[153],"mandarin":[154],"dataset":[155],"AISHELL-1.":[156],"show":[159],"regularization":[163],"indeed":[165],"help":[166],"learn":[170],"alignments.":[174],"Besides,":[175],"be":[181],"speeded":[182],"up":[183],"4":[185],"times":[186],"little":[190],"performance":[191],"degradation.":[192]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2024,"cited_by_count":2},{"year":2023,"cited_by_count":7}],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-10T00:00:00"}