{"id":"https://openalex.org/W4372342984","doi":"https://doi.org/10.1109/icassp49357.2023.10095544","title":"Neural Transducer Training: Reduced Memory Consumption with Sample-Wise Computation","display_name":"Neural Transducer Training: Reduced Memory Consumption with Sample-Wise Computation","publication_year":2023,"publication_date":"2023-05-05","ids":{"openalex":"https://openalex.org/W4372342984","doi":"https://doi.org/10.1109/icassp49357.2023.10095544"},"language":"en","primary_location":{"id":"doi:10.1109/icassp49357.2023.10095544","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1109/icassp49357.2023.10095544","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2023 - 2023 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/A5103176031","display_name":"Stefan Braun","orcid":"https://orcid.org/0000-0001-7117-0603"},"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":"Stefan Braun","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5030109648","display_name":"Erik McDermott","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":"Erik McDermott","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5033764781","display_name":"Roger Hsiao","orcid":"https://orcid.org/0000-0001-8638-4508"},"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":"Roger Hsiao","raw_affiliation_strings":["Apple"],"affiliations":[{"raw_affiliation_string":"Apple","institution_ids":["https://openalex.org/I4210107260"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5103176031"],"corresponding_institution_ids":["https://openalex.org/I4210107260"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.04075544,"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":"5"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.9998999834060669,"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":0.9998999834060669,"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":0.9995999932289124,"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":0.9990000128746033,"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/transducer","display_name":"Transducer","score":0.791713535785675},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7796177268028259},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.6527465581893921},{"id":"https://openalex.org/keywords/sample","display_name":"Sample (material)","score":0.5973796844482422},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.513750433921814},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.503074586391449},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.4723697304725647},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.4265501797199249},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.2943875193595886},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.29202479124069214},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.27764296531677246},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.0798497200012207},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.07705467939376831}],"concepts":[{"id":"https://openalex.org/C56318395","wikidata":"https://www.wikidata.org/wiki/Q215928","display_name":"Transducer","level":2,"score":0.791713535785675},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7796177268028259},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.6527465581893921},{"id":"https://openalex.org/C198531522","wikidata":"https://www.wikidata.org/wiki/Q485146","display_name":"Sample (material)","level":2,"score":0.5973796844482422},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.513750433921814},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.503074586391449},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.4723697304725647},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.4265501797199249},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2943875193595886},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.29202479124069214},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.27764296531677246},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.0798497200012207},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.07705467939376831},{"id":"https://openalex.org/C185592680","wikidata":"https://www.wikidata.org/wiki/Q2329","display_name":"Chemistry","level":0,"score":0.0},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.0},{"id":"https://openalex.org/C43617362","wikidata":"https://www.wikidata.org/wiki/Q170050","display_name":"Chromatography","level":1,"score":0.0},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icassp49357.2023.10095544","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1109/icassp49357.2023.10095544","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","score":0.4399999976158142,"display_name":"Quality Education"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":21,"referenced_works":["https://openalex.org/W854541894","https://openalex.org/W1828163288","https://openalex.org/W2127141656","https://openalex.org/W2327501763","https://openalex.org/W2911544293","https://openalex.org/W2962760690","https://openalex.org/W2962780374","https://openalex.org/W2963414781","https://openalex.org/W2963827914","https://openalex.org/W2974231335","https://openalex.org/W3007227084","https://openalex.org/W3016010032","https://openalex.org/W3146505093","https://openalex.org/W3208743843","https://openalex.org/W4283700324","https://openalex.org/W6623517193","https://openalex.org/W6638749077","https://openalex.org/W6741807409","https://openalex.org/W6747398299","https://openalex.org/W6767671539","https://openalex.org/W6802609832"],"related_works":["https://openalex.org/W2058965144","https://openalex.org/W2164382479","https://openalex.org/W2146343568","https://openalex.org/W98480971","https://openalex.org/W2150291671","https://openalex.org/W2013643406","https://openalex.org/W2027972911","https://openalex.org/W2157978810","https://openalex.org/W2597809628","https://openalex.org/W3046370962"],"abstract_inverted_index":{"The":[0],"neural":[1],"transducer":[2,56,68,124],"is":[3,15],"an":[4],"end-to-end":[5],"model":[6,14],"for":[7,17,128],"automatic":[8],"speech":[9],"recognition":[10],"(ASR).":[11],"While":[12],"the":[13,20,27,33,48,67,80,85,112,123],"well-suited":[16],"streaming":[18],"ASR,":[19],"training":[21,57,63],"process":[22],"remains":[23],"challenging.":[24],"During":[25],"training,":[26],"memory":[28,102],"requirements":[29],"may":[30],"quickly":[31],"exceed":[32],"capacity":[34],"of":[35,53,84,91,132,137,144],"state-of-the-art":[36],"GPUs,":[37],"limiting":[38],"batch":[39,130],"size":[40,131],"and":[41,50,70,82,104,126,134],"sequence":[42],"lengths.In":[43],"this":[44],"work,":[45],"we":[46,94,119],"analyze":[47],"time":[49],"space":[51],"complexity":[52],"a":[54,61,89,117,129],"typical":[55],"setup.":[58],"We":[59,75],"propose":[60],"memory-efficient":[62],"method":[64,99],"that":[65,96],"computes":[66],"loss":[69,125],"gradients":[71,127],"sample":[72],"by":[73],"sample.":[74],"present":[76],"optimizations":[77],"to":[78,111,121],"increase":[79],"efficiency":[81],"parallelism":[83],"sample-wise":[86,98],"method.":[87],"In":[88],"set":[90],"thorough":[92],"benchmarks,":[93],"show":[95],"our":[97],"significantly":[100],"reduces":[101],"usage,":[103],"performs":[105],"at":[106],"competitive":[107],"speed":[108],"when":[109],"compared":[110],"default":[113],"batched":[114],"computation.":[115],"As":[116],"highlight,":[118],"manage":[120],"compute":[122],"1024,":[133],"audio":[135],"length":[136],"40":[138],"seconds,":[139],"using":[140],"only":[141],"6":[142],"GB":[143],"memory.":[145]},"counts_by_year":[],"updated_date":"2025-12-24T23:09:58.560324","created_date":"2025-10-10T00:00:00"}