A Multi-dimensional Deep Structured State Space Approach to Speech Enhancement Using Small-footprint Models

We propose a multi-dimensional structured state space (S4) approach to speech enhancement. To better capture the spectral dependencies across the frequency axis, we focus on modifying the multi-dimensional S4 layer with whitening transformation to build new small-footprint models that also achieve g...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2023-06
Main Authors: Ku, Pin-Jui, Chao-Han, Huck Yang, Sabato Marco Siniscalchi, Chin-Hui, Lee
Format: Article
Language:English
Subjects:
Data augmentation
Domains
Speech processing
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
cited_by
cites
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Ku, Pin-Jui
Chao-Han, Huck Yang
Sabato Marco Siniscalchi
Chin-Hui, Lee
description We propose a multi-dimensional structured state space (S4) approach to speech enhancement. To better capture the spectral dependencies across the frequency axis, we focus on modifying the multi-dimensional S4 layer with whitening transformation to build new small-footprint models that also achieve good performance. We explore several S4-based deep architectures in time (T) and time-frequency (TF) domains. The 2-D S4 layer can be considered a particular convolutional layer with an infinite receptive field although it utilizes fewer parameters than a conventional convolutional layer. Evaluated on the VoiceBank-DEMAND data set, when compared with the conventional U-net model based on convolutional layers, the proposed TF-domain S4-based model is 78.6% smaller in size, yet it still achieves competitive results with a PESQ score of 3.15 with data augmentation. By increasing the model size, we can even reach a PESQ score of 3.18.
doi_str_mv 10.48550/arxiv.2306.00331
format article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2821737229</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2821737229</sourcerecordid><originalsourceid>FETCH-LOGICAL-a951-838e7c70348aa8600af53f4342f12a831eb919f8bb7883e7988b866eb9d44423</originalsourceid><addsrcrecordid>eNotj09Lw0AUxBdBsNR-AG8LnlM3722yL8dS_0KLh-i5bJIXm5pmY3YjfnwX9DTDD2aYEeImVWtNWabu7PTTfa8BVb5WCjG9EAuIkpAGuBIr709KKcgNZBkuxOdG7uc-dEnTnXnwnRtsL--ZR1mGaa7DPHETrQ0sy9HWLDfjODlbH2VwkTBH9zAc7VBzzAf57rvhQ5Zn2_dJ61wYpy7SvWu499fisrW959W_LkX5-PC2fU52r08v280usUUWdyKxqY1CTdZSrpRtM2w1amhTsIQpV0VatFRVhgjZFEQV5XmkjdYacClu_1rjzq-ZfTic3DzFW_4ABKlBA1DgL5cgWSk</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2821737229</pqid></control><display><type>article</type><title>A Multi-dimensional Deep Structured State Space Approach to Speech Enhancement Using Small-footprint Models</title><source>Publicly Available Content Database</source><creator>Ku, Pin-Jui ; Chao-Han, Huck Yang ; Sabato Marco Siniscalchi ; Chin-Hui, Lee</creator><creatorcontrib>Ku, Pin-Jui ; Chao-Han, Huck Yang ; Sabato Marco Siniscalchi ; Chin-Hui, Lee</creatorcontrib><description>We propose a multi-dimensional structured state space (S4) approach to speech enhancement. To better capture the spectral dependencies across the frequency axis, we focus on modifying the multi-dimensional S4 layer with whitening transformation to build new small-footprint models that also achieve good performance. We explore several S4-based deep architectures in time (T) and time-frequency (TF) domains. The 2-D S4 layer can be considered a particular convolutional layer with an infinite receptive field although it utilizes fewer parameters than a conventional convolutional layer. Evaluated on the VoiceBank-DEMAND data set, when compared with the conventional U-net model based on convolutional layers, the proposed TF-domain S4-based model is 78.6% smaller in size, yet it still achieves competitive results with a PESQ score of 3.15 with data augmentation. By increasing the model size, we can even reach a PESQ score of 3.18.</description><identifier>EISSN: 2331-8422</identifier><identifier>DOI: 10.48550/arxiv.2306.00331</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Data augmentation ; Domains ; Speech processing</subject><ispartof>arXiv.org, 2023-06</ispartof><rights>2023. This work is published under http://creativecommons.org/licenses/by-nc-nd/4.0/ (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>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://www.proquest.com/docview/2821737229?pq-origsite=primo$$EHTML$$P50$$Gproquest$$Hfree_for_read</linktohtml><link.rule.ids>780,784,25752,27924,37011,44589</link.rule.ids></links><search><creatorcontrib>Ku, Pin-Jui</creatorcontrib><creatorcontrib>Chao-Han, Huck Yang</creatorcontrib><creatorcontrib>Sabato Marco Siniscalchi</creatorcontrib><creatorcontrib>Chin-Hui, Lee</creatorcontrib><title>A Multi-dimensional Deep Structured State Space Approach to Speech Enhancement Using Small-footprint Models</title><title>arXiv.org</title><description>We propose a multi-dimensional structured state space (S4) approach to speech enhancement. To better capture the spectral dependencies across the frequency axis, we focus on modifying the multi-dimensional S4 layer with whitening transformation to build new small-footprint models that also achieve good performance. We explore several S4-based deep architectures in time (T) and time-frequency (TF) domains. The 2-D S4 layer can be considered a particular convolutional layer with an infinite receptive field although it utilizes fewer parameters than a conventional convolutional layer. Evaluated on the VoiceBank-DEMAND data set, when compared with the conventional U-net model based on convolutional layers, the proposed TF-domain S4-based model is 78.6% smaller in size, yet it still achieves competitive results with a PESQ score of 3.15 with data augmentation. By increasing the model size, we can even reach a PESQ score of 3.18.</description><subject>Data augmentation</subject><subject>Domains</subject><subject>Speech processing</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>PIMPY</sourceid><recordid>eNotj09Lw0AUxBdBsNR-AG8LnlM3722yL8dS_0KLh-i5bJIXm5pmY3YjfnwX9DTDD2aYEeImVWtNWabu7PTTfa8BVb5WCjG9EAuIkpAGuBIr709KKcgNZBkuxOdG7uc-dEnTnXnwnRtsL--ZR1mGaa7DPHETrQ0sy9HWLDfjODlbH2VwkTBH9zAc7VBzzAf57rvhQ5Zn2_dJ61wYpy7SvWu499fisrW959W_LkX5-PC2fU52r08v280usUUWdyKxqY1CTdZSrpRtM2w1amhTsIQpV0VatFRVhgjZFEQV5XmkjdYacClu_1rjzq-ZfTic3DzFW_4ABKlBA1DgL5cgWSk</recordid><startdate>20230601</startdate><enddate>20230601</enddate><creator>Ku, Pin-Jui</creator><creator>Chao-Han, Huck Yang</creator><creator>Sabato Marco Siniscalchi</creator><creator>Chin-Hui, Lee</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>PTHSS</scope></search><sort><creationdate>20230601</creationdate><title>A Multi-dimensional Deep Structured State Space Approach to Speech Enhancement Using Small-footprint Models</title><author>Ku, Pin-Jui ; Chao-Han, Huck Yang ; Sabato Marco Siniscalchi ; Chin-Hui, Lee</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a951-838e7c70348aa8600af53f4342f12a831eb919f8bb7883e7988b866eb9d44423</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Data augmentation</topic><topic>Domains</topic><topic>Speech processing</topic><toplevel>online_resources</toplevel><creatorcontrib>Ku, Pin-Jui</creatorcontrib><creatorcontrib>Chao-Han, Huck Yang</creatorcontrib><creatorcontrib>Sabato Marco Siniscalchi</creatorcontrib><creatorcontrib>Chin-Hui, Lee</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</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>Engineering Collection</collection><jtitle>arXiv.org</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Ku, Pin-Jui</au><au>Chao-Han, Huck Yang</au><au>Sabato Marco Siniscalchi</au><au>Chin-Hui, Lee</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>A Multi-dimensional Deep Structured State Space Approach to Speech Enhancement Using Small-footprint Models</atitle><jtitle>arXiv.org</jtitle><date>2023-06-01</date><risdate>2023</risdate><eissn>2331-8422</eissn><abstract>We propose a multi-dimensional structured state space (S4) approach to speech enhancement. To better capture the spectral dependencies across the frequency axis, we focus on modifying the multi-dimensional S4 layer with whitening transformation to build new small-footprint models that also achieve good performance. We explore several S4-based deep architectures in time (T) and time-frequency (TF) domains. The 2-D S4 layer can be considered a particular convolutional layer with an infinite receptive field although it utilizes fewer parameters than a conventional convolutional layer. Evaluated on the VoiceBank-DEMAND data set, when compared with the conventional U-net model based on convolutional layers, the proposed TF-domain S4-based model is 78.6% smaller in size, yet it still achieves competitive results with a PESQ score of 3.15 with data augmentation. By increasing the model size, we can even reach a PESQ score of 3.18.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><doi>10.48550/arxiv.2306.00331</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2023-06
issn 2331-8422
language eng
recordid cdi_proquest_journals_2821737229
source Publicly Available Content Database
subjects Data augmentation
Domains
Speech processing
title A Multi-dimensional Deep Structured State Space Approach to Speech Enhancement Using Small-footprint Models
url http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-11T01%3A04%3A09IST&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:journal&rft.genre=article&rft.atitle=A%20Multi-dimensional%20Deep%20Structured%20State%20Space%20Approach%20to%20Speech%20Enhancement%20Using%20Small-footprint%20Models&rft.jtitle=arXiv.org&rft.au=Ku,%20Pin-Jui&rft.date=2023-06-01&rft.eissn=2331-8422&rft_id=info:doi/10.48550/arxiv.2306.00331&rft_dat=%3Cproquest%3E2821737229%3C/proquest%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-a951-838e7c70348aa8600af53f4342f12a831eb919f8bb7883e7988b866eb9d44423%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=2821737229&rft_id=info:pmid/&rfr_iscdi=true