Block-Online Multi-Channel Speech Enhancement Using DNN-Supported Relative Transfer Function Estimates

This work addresses the problem of block-online processing for multi-channel speech enhancement. Such processing is vital in scenarios with moving speakers and/or when very short utterances are processed, e.g., in voice assistant scenarios. We consider several variants of a system that performs beam...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2019-12
Hauptverfasser: Malek, Jiri, Koldovsky, Zbynek, Bohac, Marek
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 Malek, Jiri
Koldovsky, Zbynek
Bohac, Marek
description This work addresses the problem of block-online processing for multi-channel speech enhancement. Such processing is vital in scenarios with moving speakers and/or when very short utterances are processed, e.g., in voice assistant scenarios. We consider several variants of a system that performs beamforming supported by DNN-based voice activity detection (VAD) followed by post-filtering. The speaker is targeted through estimating relative transfer functions between microphones. Each block of the input signals is processed independently in order to make the method applicable in highly dynamic environments. Owing to the short length of the processed block, the statistics required by the beamformer are estimated less precisely. The influence of this inaccuracy is studied and compared to the processing regime when recordings are treated as one block (batch processing). The experimental evaluation of the proposed method is performed on large datasets of CHiME-4 and on another dataset featuring moving target speaker. The experiments are evaluated in terms of objective and perceptual criteria (such as signal-to-interference ratio (SIR) or perceptual evaluation of speech quality (PESQ), respectively). Moreover, word error rate (WER) achieved by a baseline automatic speech recognition system is evaluated, for which the enhancement method serves as a front-end solution. The results indicate that the proposed method is robust with respect to short length of the processed block. Significant improvements in terms of the criteria and WER are observed even for the block length of 250 ms.
doi_str_mv 10.48550/arxiv.1905.03632
format Article
fullrecord <record><control><sourceid>proquest_arxiv</sourceid><recordid>TN_cdi_arxiv_primary_1905_03632</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2222822173</sourcerecordid><originalsourceid>FETCH-LOGICAL-a523-c1481ec56aa9c8da3fbb6921d20490d20ffdcb03b32d5735e1f5a159e712c9793</originalsourceid><addsrcrecordid>eNotkF9LwzAUxYMgOOY-gE8GfO7Mn6ZtHnVuKswN3HwuaXrjMru0JunQb2-dnod7OHC49_JD6IqSaVoIQW6V_7LHKZVETAnPODtDI8Y5TYqUsQs0CWFPCGFZzoTgI2Tum1Z_JGvXWAf4pW-iTWY75Rw0eNMB6B2euyFrOICL-C1Y944fVqtk03dd6yPU-BUaFe0R8NYrFwx4vOidjrZ1eB6iPagI4RKdG9UEmPz7GG0X8-3sKVmuH59nd8tECcYTTdOCghaZUlIXteKmqjLJaM1IKskwjal1RXjFWS1yLoAaoaiQkFOmZS75GF3_rT1BKDs_XPff5S-M8gRjaNz8NTrffvYQYrlve--Gn0o2qGCM5pz_AJ7wYn0</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2222822173</pqid></control><display><type>article</type><title>Block-Online Multi-Channel Speech Enhancement Using DNN-Supported Relative Transfer Function Estimates</title><source>arXiv.org</source><source>Free E- Journals</source><creator>Malek, Jiri ; Koldovsky, Zbynek ; Bohac, Marek</creator><creatorcontrib>Malek, Jiri ; Koldovsky, Zbynek ; Bohac, Marek</creatorcontrib><description>This work addresses the problem of block-online processing for multi-channel speech enhancement. Such processing is vital in scenarios with moving speakers and/or when very short utterances are processed, e.g., in voice assistant scenarios. We consider several variants of a system that performs beamforming supported by DNN-based voice activity detection (VAD) followed by post-filtering. The speaker is targeted through estimating relative transfer functions between microphones. Each block of the input signals is processed independently in order to make the method applicable in highly dynamic environments. Owing to the short length of the processed block, the statistics required by the beamformer are estimated less precisely. The influence of this inaccuracy is studied and compared to the processing regime when recordings are treated as one block (batch processing). The experimental evaluation of the proposed method is performed on large datasets of CHiME-4 and on another dataset featuring moving target speaker. The experiments are evaluated in terms of objective and perceptual criteria (such as signal-to-interference ratio (SIR) or perceptual evaluation of speech quality (PESQ), respectively). Moreover, word error rate (WER) achieved by a baseline automatic speech recognition system is evaluated, for which the enhancement method serves as a front-end solution. The results indicate that the proposed method is robust with respect to short length of the processed block. Significant improvements in terms of the criteria and WER are observed even for the block length of 250 ms.</description><identifier>EISSN: 2331-8422</identifier><identifier>DOI: 10.48550/arxiv.1905.03632</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Automatic speech recognition ; Batch processing ; Beamforming ; Computer Science - Sound ; Computer Science - Systems and Control ; Microphones ; Robustness (mathematics) ; Signal processing ; Speech processing ; Transfer functions ; Voice activity detectors ; Voice recognition</subject><ispartof>arXiv.org, 2019-12</ispartof><rights>2019. 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><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,784,885,27925</link.rule.ids><backlink>$$Uhttps://doi.org/10.1049/iet-spr.2019.0304$$DView published paper (Access to full text may be restricted)$$Hfree_for_read</backlink><backlink>$$Uhttps://doi.org/10.48550/arXiv.1905.03632$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Malek, Jiri</creatorcontrib><creatorcontrib>Koldovsky, Zbynek</creatorcontrib><creatorcontrib>Bohac, Marek</creatorcontrib><title>Block-Online Multi-Channel Speech Enhancement Using DNN-Supported Relative Transfer Function Estimates</title><title>arXiv.org</title><description>This work addresses the problem of block-online processing for multi-channel speech enhancement. Such processing is vital in scenarios with moving speakers and/or when very short utterances are processed, e.g., in voice assistant scenarios. We consider several variants of a system that performs beamforming supported by DNN-based voice activity detection (VAD) followed by post-filtering. The speaker is targeted through estimating relative transfer functions between microphones. Each block of the input signals is processed independently in order to make the method applicable in highly dynamic environments. Owing to the short length of the processed block, the statistics required by the beamformer are estimated less precisely. The influence of this inaccuracy is studied and compared to the processing regime when recordings are treated as one block (batch processing). The experimental evaluation of the proposed method is performed on large datasets of CHiME-4 and on another dataset featuring moving target speaker. The experiments are evaluated in terms of objective and perceptual criteria (such as signal-to-interference ratio (SIR) or perceptual evaluation of speech quality (PESQ), respectively). Moreover, word error rate (WER) achieved by a baseline automatic speech recognition system is evaluated, for which the enhancement method serves as a front-end solution. The results indicate that the proposed method is robust with respect to short length of the processed block. Significant improvements in terms of the criteria and WER are observed even for the block length of 250 ms.</description><subject>Automatic speech recognition</subject><subject>Batch processing</subject><subject>Beamforming</subject><subject>Computer Science - Sound</subject><subject>Computer Science - Systems and Control</subject><subject>Microphones</subject><subject>Robustness (mathematics)</subject><subject>Signal processing</subject><subject>Speech processing</subject><subject>Transfer functions</subject><subject>Voice activity detectors</subject><subject>Voice recognition</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2019</creationdate><recordtype>article</recordtype><sourceid>ABUWG</sourceid><sourceid>AFKRA</sourceid><sourceid>AZQEC</sourceid><sourceid>BENPR</sourceid><sourceid>CCPQU</sourceid><sourceid>DWQXO</sourceid><sourceid>GOX</sourceid><recordid>eNotkF9LwzAUxYMgOOY-gE8GfO7Mn6ZtHnVuKswN3HwuaXrjMru0JunQb2-dnod7OHC49_JD6IqSaVoIQW6V_7LHKZVETAnPODtDI8Y5TYqUsQs0CWFPCGFZzoTgI2Tum1Z_JGvXWAf4pW-iTWY75Rw0eNMB6B2euyFrOICL-C1Y944fVqtk03dd6yPU-BUaFe0R8NYrFwx4vOidjrZ1eB6iPagI4RKdG9UEmPz7GG0X8-3sKVmuH59nd8tECcYTTdOCghaZUlIXteKmqjLJaM1IKskwjal1RXjFWS1yLoAaoaiQkFOmZS75GF3_rT1BKDs_XPff5S-M8gRjaNz8NTrffvYQYrlve--Gn0o2qGCM5pz_AJ7wYn0</recordid><startdate>20191211</startdate><enddate>20191211</enddate><creator>Malek, Jiri</creator><creator>Koldovsky, Zbynek</creator><creator>Bohac, Marek</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><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20191211</creationdate><title>Block-Online Multi-Channel Speech Enhancement Using DNN-Supported Relative Transfer Function Estimates</title><author>Malek, Jiri ; Koldovsky, Zbynek ; Bohac, Marek</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a523-c1481ec56aa9c8da3fbb6921d20490d20ffdcb03b32d5735e1f5a159e712c9793</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Automatic speech recognition</topic><topic>Batch processing</topic><topic>Beamforming</topic><topic>Computer Science - Sound</topic><topic>Computer Science - Systems and Control</topic><topic>Microphones</topic><topic>Robustness (mathematics)</topic><topic>Signal processing</topic><topic>Speech processing</topic><topic>Transfer functions</topic><topic>Voice activity detectors</topic><topic>Voice recognition</topic><toplevel>online_resources</toplevel><creatorcontrib>Malek, Jiri</creatorcontrib><creatorcontrib>Koldovsky, Zbynek</creatorcontrib><creatorcontrib>Bohac, Marek</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><collection>arXiv Computer Science</collection><collection>arXiv.org</collection><jtitle>arXiv.org</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Malek, Jiri</au><au>Koldovsky, Zbynek</au><au>Bohac, Marek</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Block-Online Multi-Channel Speech Enhancement Using DNN-Supported Relative Transfer Function Estimates</atitle><jtitle>arXiv.org</jtitle><date>2019-12-11</date><risdate>2019</risdate><eissn>2331-8422</eissn><abstract>This work addresses the problem of block-online processing for multi-channel speech enhancement. Such processing is vital in scenarios with moving speakers and/or when very short utterances are processed, e.g., in voice assistant scenarios. We consider several variants of a system that performs beamforming supported by DNN-based voice activity detection (VAD) followed by post-filtering. The speaker is targeted through estimating relative transfer functions between microphones. Each block of the input signals is processed independently in order to make the method applicable in highly dynamic environments. Owing to the short length of the processed block, the statistics required by the beamformer are estimated less precisely. The influence of this inaccuracy is studied and compared to the processing regime when recordings are treated as one block (batch processing). The experimental evaluation of the proposed method is performed on large datasets of CHiME-4 and on another dataset featuring moving target speaker. The experiments are evaluated in terms of objective and perceptual criteria (such as signal-to-interference ratio (SIR) or perceptual evaluation of speech quality (PESQ), respectively). Moreover, word error rate (WER) achieved by a baseline automatic speech recognition system is evaluated, for which the enhancement method serves as a front-end solution. The results indicate that the proposed method is robust with respect to short length of the processed block. Significant improvements in terms of the criteria and WER are observed even for the block length of 250 ms.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><doi>10.48550/arxiv.1905.03632</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2019-12
issn 2331-8422
language eng
recordid cdi_arxiv_primary_1905_03632
source arXiv.org; Free E- Journals
subjects Automatic speech recognition
Batch processing
Beamforming
Computer Science - Sound
Computer Science - Systems and Control
Microphones
Robustness (mathematics)
Signal processing
Speech processing
Transfer functions
Voice activity detectors
Voice recognition
title Block-Online Multi-Channel Speech Enhancement Using DNN-Supported Relative Transfer Function Estimates
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-23T09%3A02%3A27IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest_arxiv&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Block-Online%20Multi-Channel%20Speech%20Enhancement%20Using%20DNN-Supported%20Relative%20Transfer%20Function%20Estimates&rft.jtitle=arXiv.org&rft.au=Malek,%20Jiri&rft.date=2019-12-11&rft.eissn=2331-8422&rft_id=info:doi/10.48550/arxiv.1905.03632&rft_dat=%3Cproquest_arxiv%3E2222822173%3C/proquest_arxiv%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2222822173&rft_id=info:pmid/&rfr_iscdi=true