Loading…
A Novel Mask Estimation Method Employing Posterior-Based Representative Mean Estimate for Missing-Feature Speech Recognition
This paper proposes a novel mask estimation method for missing-feature reconstruction to improve speech recognition performance in various types of background noise conditions. A conventional mask estimation method based on spectral subtraction degrades performance, due to incorrect estimation of th...
Saved in:
Published in: | IEEE transactions on audio, speech, and language processing speech, and language processing, 2011-07, Vol.19 (5), p.1434-1443 |
---|---|
Main Authors: | , |
Format: | Article |
Language: | English |
Subjects: | |
Citations: | Items that this one cites Items that cite this one |
Online Access: | Get full text |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
cited_by | cdi_FETCH-LOGICAL-c355t-b2b84daae63248877751058632599c2d059fc5468259cf3af8cade4b6c0b2dd23 |
---|---|
cites | cdi_FETCH-LOGICAL-c355t-b2b84daae63248877751058632599c2d059fc5468259cf3af8cade4b6c0b2dd23 |
container_end_page | 1443 |
container_issue | 5 |
container_start_page | 1434 |
container_title | IEEE transactions on audio, speech, and language processing |
container_volume | 19 |
creator | Wooil Kim Hansen, John H L |
description | This paper proposes a novel mask estimation method for missing-feature reconstruction to improve speech recognition performance in various types of background noise conditions. A conventional mask estimation method based on spectral subtraction degrades performance, due to incorrect estimation of the noise signal which fails to accurately represent the variations of background noise during the incoming speech utterance. The proposed mask estimation method utilizes a Posterior-based Representative Mean (PRM) estimate for determining the reliability of the input speech spectral components, which is obtained as a weighted sum of the mean parameters of the speech model using the posterior probability. To obtain the noise-corrupted speech model, a model combination method is employed, which was proposed in our previous study for a feature compensation method. Experimental results demonstrate that the proposed mask estimation method provides more separable distributions for the reliable/unreliable component classifier compared to the conventional mask estimation method. The recognition performance is evaluated using the Aurora 2.0 framework over various types of background noise conditions and the CU-Move real-life in-vehicle corpus. The performance evaluation shows that the proposed mask estimation method is considerably more effective at increasing speech recognition performance in various types of background noise conditions, compared to the conventional mask estimation method which is based on spectral subtraction. By employing the proposed PRM-based mask estimation for missing-feature reconstruction, we obtain +23.41% and +9.45% average relative improvements in word error rate for all four types of noise conditions and CU-Move corpus, respectively, compared to conventional mask estimation methods. |
doi_str_mv | 10.1109/TASL.2010.2091633 |
format | article |
fullrecord | <record><control><sourceid>proquest_pasca</sourceid><recordid>TN_cdi_pascalfrancis_primary_24286322</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>5667043</ieee_id><sourcerecordid>2559708021</sourcerecordid><originalsourceid>FETCH-LOGICAL-c355t-b2b84daae63248877751058632599c2d059fc5468259cf3af8cade4b6c0b2dd23</originalsourceid><addsrcrecordid>eNpdkU1LxDAQhoso-PkDxEsQBC9dk7Rp0-Mq6wfsqrjruaTpVKPdpmbaBcEfb8que_CUDHned2byBsEpoyPGaHa1GM-nI059yWnGkijaCQ6YEDJMMx7vbu8s2Q8OET8ojaMkZgfBz5g82hXUZKbwk0ywM0vVGduQGXTvtiSTZVvbb9O8kWeLHThjXXitEEryAq0DhKbz_Ao8r5o_PZDKOjIziF4Y3oLqegdk3gLod6_T9q0xQ5PjYK9SNcLJ5jwKXm8ni5v7cPp093AznoY6EqILC17IuFQKkojHUqZpKhgV0lciyzQvqcgqLeJE-lpXkaqkViXERaJpwcuSR0fB5dq3dfarB-zypUENda0asD3mzP-blNQ7e_T8H_phe9f46fKMiSRJpWQeYmtIO4vooMpb5_d2394pH-LIhzjyIY58E4fXXGyMFWpVV0412uBWyGM-LDTMerbmDABsn4fOPrLoF6jqk9Y</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>915667881</pqid></control><display><type>article</type><title>A Novel Mask Estimation Method Employing Posterior-Based Representative Mean Estimate for Missing-Feature Speech Recognition</title><source>IEEE Xplore (Online service)</source><creator>Wooil Kim ; Hansen, John H L</creator><creatorcontrib>Wooil Kim ; Hansen, John H L</creatorcontrib><description>This paper proposes a novel mask estimation method for missing-feature reconstruction to improve speech recognition performance in various types of background noise conditions. A conventional mask estimation method based on spectral subtraction degrades performance, due to incorrect estimation of the noise signal which fails to accurately represent the variations of background noise during the incoming speech utterance. The proposed mask estimation method utilizes a Posterior-based Representative Mean (PRM) estimate for determining the reliability of the input speech spectral components, which is obtained as a weighted sum of the mean parameters of the speech model using the posterior probability. To obtain the noise-corrupted speech model, a model combination method is employed, which was proposed in our previous study for a feature compensation method. Experimental results demonstrate that the proposed mask estimation method provides more separable distributions for the reliable/unreliable component classifier compared to the conventional mask estimation method. The recognition performance is evaluated using the Aurora 2.0 framework over various types of background noise conditions and the CU-Move real-life in-vehicle corpus. The performance evaluation shows that the proposed mask estimation method is considerably more effective at increasing speech recognition performance in various types of background noise conditions, compared to the conventional mask estimation method which is based on spectral subtraction. By employing the proposed PRM-based mask estimation for missing-feature reconstruction, we obtain +23.41% and +9.45% average relative improvements in word error rate for all four types of noise conditions and CU-Move corpus, respectively, compared to conventional mask estimation methods.</description><identifier>ISSN: 1558-7916</identifier><identifier>ISSN: 2329-9290</identifier><identifier>EISSN: 1558-7924</identifier><identifier>EISSN: 2329-9304</identifier><identifier>DOI: 10.1109/TASL.2010.2091633</identifier><identifier>CODEN: ITASD8</identifier><language>eng</language><publisher>Piscataway, NJ: IEEE</publisher><subject>Applied sciences ; Background noise ; Estimates ; Exact sciences and technology ; Information, signal and communications theory ; mask estimation ; Masks ; Mathematical models ; missing-feature ; Noise ; Pattern recognition ; posterior-based representative mean (PRM) estimate ; Reconstruction ; robust speech recognition ; Signal and communications theory ; Signal processing ; Signal representation. Spectral analysis ; Signal, noise ; Spectra ; Speech ; Speech processing ; Speech recognition ; Studies ; Telecommunications and information theory</subject><ispartof>IEEE transactions on audio, speech, and language processing, 2011-07, Vol.19 (5), p.1434-1443</ispartof><rights>2015 INIST-CNRS</rights><rights>Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) Jul 2011</rights><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c355t-b2b84daae63248877751058632599c2d059fc5468259cf3af8cade4b6c0b2dd23</citedby><cites>FETCH-LOGICAL-c355t-b2b84daae63248877751058632599c2d059fc5468259cf3af8cade4b6c0b2dd23</cites></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/5667043$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>314,780,784,27924,27925,54796</link.rule.ids><backlink>$$Uhttp://pascal-francis.inist.fr/vibad/index.php?action=getRecordDetail&idt=24286322$$DView record in Pascal Francis$$Hfree_for_read</backlink></links><search><creatorcontrib>Wooil Kim</creatorcontrib><creatorcontrib>Hansen, John H L</creatorcontrib><title>A Novel Mask Estimation Method Employing Posterior-Based Representative Mean Estimate for Missing-Feature Speech Recognition</title><title>IEEE transactions on audio, speech, and language processing</title><addtitle>TASL</addtitle><description>This paper proposes a novel mask estimation method for missing-feature reconstruction to improve speech recognition performance in various types of background noise conditions. A conventional mask estimation method based on spectral subtraction degrades performance, due to incorrect estimation of the noise signal which fails to accurately represent the variations of background noise during the incoming speech utterance. The proposed mask estimation method utilizes a Posterior-based Representative Mean (PRM) estimate for determining the reliability of the input speech spectral components, which is obtained as a weighted sum of the mean parameters of the speech model using the posterior probability. To obtain the noise-corrupted speech model, a model combination method is employed, which was proposed in our previous study for a feature compensation method. Experimental results demonstrate that the proposed mask estimation method provides more separable distributions for the reliable/unreliable component classifier compared to the conventional mask estimation method. The recognition performance is evaluated using the Aurora 2.0 framework over various types of background noise conditions and the CU-Move real-life in-vehicle corpus. The performance evaluation shows that the proposed mask estimation method is considerably more effective at increasing speech recognition performance in various types of background noise conditions, compared to the conventional mask estimation method which is based on spectral subtraction. By employing the proposed PRM-based mask estimation for missing-feature reconstruction, we obtain +23.41% and +9.45% average relative improvements in word error rate for all four types of noise conditions and CU-Move corpus, respectively, compared to conventional mask estimation methods.</description><subject>Applied sciences</subject><subject>Background noise</subject><subject>Estimates</subject><subject>Exact sciences and technology</subject><subject>Information, signal and communications theory</subject><subject>mask estimation</subject><subject>Masks</subject><subject>Mathematical models</subject><subject>missing-feature</subject><subject>Noise</subject><subject>Pattern recognition</subject><subject>posterior-based representative mean (PRM) estimate</subject><subject>Reconstruction</subject><subject>robust speech recognition</subject><subject>Signal and communications theory</subject><subject>Signal processing</subject><subject>Signal representation. Spectral analysis</subject><subject>Signal, noise</subject><subject>Spectra</subject><subject>Speech</subject><subject>Speech processing</subject><subject>Speech recognition</subject><subject>Studies</subject><subject>Telecommunications and information theory</subject><issn>1558-7916</issn><issn>2329-9290</issn><issn>1558-7924</issn><issn>2329-9304</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2011</creationdate><recordtype>article</recordtype><recordid>eNpdkU1LxDAQhoso-PkDxEsQBC9dk7Rp0-Mq6wfsqrjruaTpVKPdpmbaBcEfb8que_CUDHned2byBsEpoyPGaHa1GM-nI059yWnGkijaCQ6YEDJMMx7vbu8s2Q8OET8ojaMkZgfBz5g82hXUZKbwk0ywM0vVGduQGXTvtiSTZVvbb9O8kWeLHThjXXitEEryAq0DhKbz_Ao8r5o_PZDKOjIziF4Y3oLqegdk3gLod6_T9q0xQ5PjYK9SNcLJ5jwKXm8ni5v7cPp093AznoY6EqILC17IuFQKkojHUqZpKhgV0lciyzQvqcgqLeJE-lpXkaqkViXERaJpwcuSR0fB5dq3dfarB-zypUENda0asD3mzP-blNQ7e_T8H_phe9f46fKMiSRJpWQeYmtIO4vooMpb5_d2394pH-LIhzjyIY58E4fXXGyMFWpVV0412uBWyGM-LDTMerbmDABsn4fOPrLoF6jqk9Y</recordid><startdate>20110701</startdate><enddate>20110701</enddate><creator>Wooil Kim</creator><creator>Hansen, John H L</creator><general>IEEE</general><general>Institute of Electrical and Electronics Engineers</general><general>The Institute of Electrical and Electronics Engineers, Inc. (IEEE)</general><scope>97E</scope><scope>RIA</scope><scope>RIE</scope><scope>IQODW</scope><scope>AAYXX</scope><scope>CITATION</scope><scope>7SC</scope><scope>8FD</scope><scope>JQ2</scope><scope>L7M</scope><scope>L~C</scope><scope>L~D</scope></search><sort><creationdate>20110701</creationdate><title>A Novel Mask Estimation Method Employing Posterior-Based Representative Mean Estimate for Missing-Feature Speech Recognition</title><author>Wooil Kim ; Hansen, John H L</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c355t-b2b84daae63248877751058632599c2d059fc5468259cf3af8cade4b6c0b2dd23</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2011</creationdate><topic>Applied sciences</topic><topic>Background noise</topic><topic>Estimates</topic><topic>Exact sciences and technology</topic><topic>Information, signal and communications theory</topic><topic>mask estimation</topic><topic>Masks</topic><topic>Mathematical models</topic><topic>missing-feature</topic><topic>Noise</topic><topic>Pattern recognition</topic><topic>posterior-based representative mean (PRM) estimate</topic><topic>Reconstruction</topic><topic>robust speech recognition</topic><topic>Signal and communications theory</topic><topic>Signal processing</topic><topic>Signal representation. Spectral analysis</topic><topic>Signal, noise</topic><topic>Spectra</topic><topic>Speech</topic><topic>Speech processing</topic><topic>Speech recognition</topic><topic>Studies</topic><topic>Telecommunications and information theory</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Wooil Kim</creatorcontrib><creatorcontrib>Hansen, John H L</creatorcontrib><collection>IEEE All-Society Periodicals Package (ASPP) 2005-present</collection><collection>IEEE All-Society Periodicals Package (ASPP) 1998–Present</collection><collection>IEEE/IET Electronic Library (IEL)</collection><collection>Pascal-Francis</collection><collection>CrossRef</collection><collection>Computer and Information Systems Abstracts</collection><collection>Technology Research Database</collection><collection>ProQuest Computer Science Collection</collection><collection>Advanced Technologies Database with Aerospace</collection><collection>Computer and Information Systems Abstracts Academic</collection><collection>Computer and Information Systems Abstracts Professional</collection><jtitle>IEEE transactions on audio, speech, and language processing</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Wooil Kim</au><au>Hansen, John H L</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>A Novel Mask Estimation Method Employing Posterior-Based Representative Mean Estimate for Missing-Feature Speech Recognition</atitle><jtitle>IEEE transactions on audio, speech, and language processing</jtitle><stitle>TASL</stitle><date>2011-07-01</date><risdate>2011</risdate><volume>19</volume><issue>5</issue><spage>1434</spage><epage>1443</epage><pages>1434-1443</pages><issn>1558-7916</issn><issn>2329-9290</issn><eissn>1558-7924</eissn><eissn>2329-9304</eissn><coden>ITASD8</coden><abstract>This paper proposes a novel mask estimation method for missing-feature reconstruction to improve speech recognition performance in various types of background noise conditions. A conventional mask estimation method based on spectral subtraction degrades performance, due to incorrect estimation of the noise signal which fails to accurately represent the variations of background noise during the incoming speech utterance. The proposed mask estimation method utilizes a Posterior-based Representative Mean (PRM) estimate for determining the reliability of the input speech spectral components, which is obtained as a weighted sum of the mean parameters of the speech model using the posterior probability. To obtain the noise-corrupted speech model, a model combination method is employed, which was proposed in our previous study for a feature compensation method. Experimental results demonstrate that the proposed mask estimation method provides more separable distributions for the reliable/unreliable component classifier compared to the conventional mask estimation method. The recognition performance is evaluated using the Aurora 2.0 framework over various types of background noise conditions and the CU-Move real-life in-vehicle corpus. The performance evaluation shows that the proposed mask estimation method is considerably more effective at increasing speech recognition performance in various types of background noise conditions, compared to the conventional mask estimation method which is based on spectral subtraction. By employing the proposed PRM-based mask estimation for missing-feature reconstruction, we obtain +23.41% and +9.45% average relative improvements in word error rate for all four types of noise conditions and CU-Move corpus, respectively, compared to conventional mask estimation methods.</abstract><cop>Piscataway, NJ</cop><pub>IEEE</pub><doi>10.1109/TASL.2010.2091633</doi><tpages>10</tpages></addata></record> |
fulltext | fulltext |
identifier | ISSN: 1558-7916 |
ispartof | IEEE transactions on audio, speech, and language processing, 2011-07, Vol.19 (5), p.1434-1443 |
issn | 1558-7916 2329-9290 1558-7924 2329-9304 |
language | eng |
recordid | cdi_pascalfrancis_primary_24286322 |
source | IEEE Xplore (Online service) |
subjects | Applied sciences Background noise Estimates Exact sciences and technology Information, signal and communications theory mask estimation Masks Mathematical models missing-feature Noise Pattern recognition posterior-based representative mean (PRM) estimate Reconstruction robust speech recognition Signal and communications theory Signal processing Signal representation. Spectral analysis Signal, noise Spectra Speech Speech processing Speech recognition Studies Telecommunications and information theory |
title | A Novel Mask Estimation Method Employing Posterior-Based Representative Mean Estimate for Missing-Feature Speech Recognition |
url | http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-24T16%3A01%3A13IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_pasca&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=A%20Novel%20Mask%20Estimation%20Method%20Employing%20Posterior-Based%20Representative%20Mean%20Estimate%20for%20Missing-Feature%20Speech%20Recognition&rft.jtitle=IEEE%20transactions%20on%20audio,%20speech,%20and%20language%20processing&rft.au=Wooil%20Kim&rft.date=2011-07-01&rft.volume=19&rft.issue=5&rft.spage=1434&rft.epage=1443&rft.pages=1434-1443&rft.issn=1558-7916&rft.eissn=1558-7924&rft.coden=ITASD8&rft_id=info:doi/10.1109/TASL.2010.2091633&rft_dat=%3Cproquest_pasca%3E2559708021%3C/proquest_pasca%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c355t-b2b84daae63248877751058632599c2d059fc5468259cf3af8cade4b6c0b2dd23%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=915667881&rft_id=info:pmid/&rft_ieee_id=5667043&rfr_iscdi=true |