Bridging the Granularity Gap for Acoustic Modeling

While Transformer has become the de-facto standard for speech, modeling upon the fine-grained frame-level features remains an open challenge of capturing long-distance dependencies and distributing the attention weights. We propose \textit{Progressive Down-Sampling} (PDS) which gradually compresses...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2023-05
Hauptverfasser: Chen, Xu, Zhang, Yuhao, Jiao, Chengbo, Liu, Xiaoqian, Hu, Chi, Zeng, Xin, Tong, Xiao, Ma, Anxiang, Wang, Huizhen, Zhu, JingBo
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Chen, Xu
Zhang, Yuhao
Jiao, Chengbo
Liu, Xiaoqian
Hu, Chi
Zeng, Xin
Tong, Xiao
Ma, Anxiang
Wang, Huizhen
Zhu, JingBo
description While Transformer has become the de-facto standard for speech, modeling upon the fine-grained frame-level features remains an open challenge of capturing long-distance dependencies and distributing the attention weights. We propose \textit{Progressive Down-Sampling} (PDS) which gradually compresses the acoustic features into coarser-grained units containing more complete semantic information, like text-level representation. In addition, we develop a representation fusion method to alleviate information loss that occurs inevitably during high compression. In this way, we compress the acoustic features into 1/32 of the initial length while achieving better or comparable performances on the speech recognition task. And as a bonus, it yields inference speedups ranging from 1.20\(\times\) to 1.47\(\times\). By reducing the modeling burden, we also achieve competitive results when training on the more challenging speech translation task.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2820821666</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2820821666</sourcerecordid><originalsourceid>FETCH-proquest_journals_28208216663</originalsourceid><addsrcrecordid>eNpjYuA0MjY21LUwMTLiYOAtLs4yMDAwMjM3MjU15mQwcirKTEnPzEtXKMlIVXAvSswrzUksyiypVHBPLFBIyy9ScEzOLy0uyUxW8M1PSc0BquRhYE1LzClO5YXS3AzKbq4hzh66BUX5haWpxSXxWfmlRXlAqXgjCyMDCyNDMzMzY-JUAQC6zDOL</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2820821666</pqid></control><display><type>article</type><title>Bridging the Granularity Gap for Acoustic Modeling</title><source>Free E- Journals</source><creator>Chen, Xu ; Zhang, Yuhao ; Jiao, Chengbo ; Liu, Xiaoqian ; Hu, Chi ; Zeng, Xin ; Tong, Xiao ; Ma, Anxiang ; Wang, Huizhen ; Zhu, JingBo</creator><creatorcontrib>Chen, Xu ; Zhang, Yuhao ; Jiao, Chengbo ; Liu, Xiaoqian ; Hu, Chi ; Zeng, Xin ; Tong, Xiao ; Ma, Anxiang ; Wang, Huizhen ; Zhu, JingBo</creatorcontrib><description>While Transformer has become the de-facto standard for speech, modeling upon the fine-grained frame-level features remains an open challenge of capturing long-distance dependencies and distributing the attention weights. We propose \textit{Progressive Down-Sampling} (PDS) which gradually compresses the acoustic features into coarser-grained units containing more complete semantic information, like text-level representation. In addition, we develop a representation fusion method to alleviate information loss that occurs inevitably during high compression. In this way, we compress the acoustic features into 1/32 of the initial length while achieving better or comparable performances on the speech recognition task. And as a bonus, it yields inference speedups ranging from 1.20\(\times\) to 1.47\(\times\). By reducing the modeling burden, we also achieve competitive results when training on the more challenging speech translation task.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Modelling ; Representations ; Speech recognition</subject><ispartof>arXiv.org, 2023-05</ispartof><rights>2023. This work is published under http://arxiv.org/licenses/nonexclusive-distrib/1.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>778,782</link.rule.ids></links><search><creatorcontrib>Chen, Xu</creatorcontrib><creatorcontrib>Zhang, Yuhao</creatorcontrib><creatorcontrib>Jiao, Chengbo</creatorcontrib><creatorcontrib>Liu, Xiaoqian</creatorcontrib><creatorcontrib>Hu, Chi</creatorcontrib><creatorcontrib>Zeng, Xin</creatorcontrib><creatorcontrib>Tong, Xiao</creatorcontrib><creatorcontrib>Ma, Anxiang</creatorcontrib><creatorcontrib>Wang, Huizhen</creatorcontrib><creatorcontrib>Zhu, JingBo</creatorcontrib><title>Bridging the Granularity Gap for Acoustic Modeling</title><title>arXiv.org</title><description>While Transformer has become the de-facto standard for speech, modeling upon the fine-grained frame-level features remains an open challenge of capturing long-distance dependencies and distributing the attention weights. We propose \textit{Progressive Down-Sampling} (PDS) which gradually compresses the acoustic features into coarser-grained units containing more complete semantic information, like text-level representation. In addition, we develop a representation fusion method to alleviate information loss that occurs inevitably during high compression. In this way, we compress the acoustic features into 1/32 of the initial length while achieving better or comparable performances on the speech recognition task. And as a bonus, it yields inference speedups ranging from 1.20\(\times\) to 1.47\(\times\). By reducing the modeling burden, we also achieve competitive results when training on the more challenging speech translation task.</description><subject>Modelling</subject><subject>Representations</subject><subject>Speech recognition</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>ABUWG</sourceid><sourceid>AFKRA</sourceid><sourceid>AZQEC</sourceid><sourceid>BENPR</sourceid><sourceid>CCPQU</sourceid><sourceid>DWQXO</sourceid><recordid>eNpjYuA0MjY21LUwMTLiYOAtLs4yMDAwMjM3MjU15mQwcirKTEnPzEtXKMlIVXAvSswrzUksyiypVHBPLFBIyy9ScEzOLy0uyUxW8M1PSc0BquRhYE1LzClO5YXS3AzKbq4hzh66BUX5haWpxSXxWfmlRXlAqXgjCyMDCyNDMzMzY-JUAQC6zDOL</recordid><startdate>20230527</startdate><enddate>20230527</enddate><creator>Chen, Xu</creator><creator>Zhang, Yuhao</creator><creator>Jiao, Chengbo</creator><creator>Liu, Xiaoqian</creator><creator>Hu, Chi</creator><creator>Zeng, Xin</creator><creator>Tong, Xiao</creator><creator>Ma, Anxiang</creator><creator>Wang, Huizhen</creator><creator>Zhu, JingBo</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope></search><sort><creationdate>20230527</creationdate><title>Bridging the Granularity Gap for Acoustic Modeling</title><author>Chen, Xu ; Zhang, Yuhao ; Jiao, Chengbo ; Liu, Xiaoqian ; Hu, Chi ; Zeng, Xin ; Tong, Xiao ; Ma, Anxiang ; Wang, Huizhen ; Zhu, JingBo</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_28208216663</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Modelling</topic><topic>Representations</topic><topic>Speech recognition</topic><toplevel>online_resources</toplevel><creatorcontrib>Chen, Xu</creatorcontrib><creatorcontrib>Zhang, Yuhao</creatorcontrib><creatorcontrib>Jiao, Chengbo</creatorcontrib><creatorcontrib>Liu, Xiaoqian</creatorcontrib><creatorcontrib>Hu, Chi</creatorcontrib><creatorcontrib>Zeng, Xin</creatorcontrib><creatorcontrib>Tong, Xiao</creatorcontrib><creatorcontrib>Ma, Anxiang</creatorcontrib><creatorcontrib>Wang, Huizhen</creatorcontrib><creatorcontrib>Zhu, JingBo</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni Edition)</collection><collection>ProQuest Central UK/Ireland</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central Korea</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Chen, Xu</au><au>Zhang, Yuhao</au><au>Jiao, Chengbo</au><au>Liu, Xiaoqian</au><au>Hu, Chi</au><au>Zeng, Xin</au><au>Tong, Xiao</au><au>Ma, Anxiang</au><au>Wang, Huizhen</au><au>Zhu, JingBo</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Bridging the Granularity Gap for Acoustic Modeling</atitle><jtitle>arXiv.org</jtitle><date>2023-05-27</date><risdate>2023</risdate><eissn>2331-8422</eissn><abstract>While Transformer has become the de-facto standard for speech, modeling upon the fine-grained frame-level features remains an open challenge of capturing long-distance dependencies and distributing the attention weights. We propose \textit{Progressive Down-Sampling} (PDS) which gradually compresses the acoustic features into coarser-grained units containing more complete semantic information, like text-level representation. In addition, we develop a representation fusion method to alleviate information loss that occurs inevitably during high compression. In this way, we compress the acoustic features into 1/32 of the initial length while achieving better or comparable performances on the speech recognition task. And as a bonus, it yields inference speedups ranging from 1.20\(\times\) to 1.47\(\times\). By reducing the modeling burden, we also achieve competitive results when training on the more challenging speech translation task.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2023-05
issn 2331-8422
language eng
recordid cdi_proquest_journals_2820821666
source Free E- Journals
subjects Modelling
Representations
Speech recognition
title Bridging the Granularity Gap for Acoustic Modeling
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-16T22%3A15%3A22IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=Bridging%20the%20Granularity%20Gap%20for%20Acoustic%20Modeling&rft.jtitle=arXiv.org&rft.au=Chen,%20Xu&rft.date=2023-05-27&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2820821666%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2820821666&rft_id=info:pmid/&rfr_iscdi=true