Loading…
Multi-Scale multi-band densenets for audio source separation
This paper deals with the problem of audio source separation. To handle the complex and ill-posed nature of the problems of audio source separation, the current state-of-the-art approaches employ deep neural networks to obtain instrumental spectra from a mixture. In this study, we propose a novel ne...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Conference Proceeding |
| Language: | English |
| Subjects: | |
| Citations: | Items that cite this one |
| Online Access: | Request full text |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| cited_by | cdi_FETCH-LOGICAL-c225t-d5b37b235669aeab037619743ae32ad361ba2c25aa73148818eeed953415846c3 |
|---|---|
| cites | |
| container_end_page | 25 |
| container_issue | |
| container_start_page | 21 |
| container_title | |
| container_volume | |
| creator | Takahashi, Naoya Mitsufuji, Yuki |
| description | This paper deals with the problem of audio source separation. To handle the complex and ill-posed nature of the problems of audio source separation, the current state-of-the-art approaches employ deep neural networks to obtain instrumental spectra from a mixture. In this study, we propose a novel network architecture that extends the recently developed densely connected convolutional network (DenseNet), which has shown excellent results on image classification tasks. To deal with the specific problem of audio source separation, an up-sampling layer, block skip connection and band-dedicated dense blocks are incorporated on top of DenseNet. The proposed approach takes advantage of long contextual information and outperforms state-of-the-art results on SiSEC 2016 competition by a large margin in terms of signal-to-distortion ratio. Moreover, the proposed architecture requires significantly fewer parameters and considerably less training time compared with other methods. |
| doi_str_mv | 10.1109/WASPAA.2017.8169987 |
| format | conference_proceeding |
| fullrecord | <record><control><sourceid>ieee_CHZPO</sourceid><recordid>TN_cdi_ieee_primary_8169987</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><ieee_id>8169987</ieee_id><sourcerecordid>8169987</sourcerecordid><originalsourceid>FETCH-LOGICAL-c225t-d5b37b235669aeab037619743ae32ad361ba2c25aa73148818eeed953415846c3</originalsourceid><addsrcrecordid>eNotj81Kw0AURkdBsNY-QTfzAqlz584vuAlFrVBRqOKy3GRuIZImJZMsfHtFu_oOHDjwCbEEtQJQ8e6z3L2V5Uor8KsALsbgL8Qi-gAWgwOHGi7FDKLxBTgdr8VNzl9KWR2Mmon7l6kdm2JXU8vy-McVdUkm7jJ3PGZ56AdJU2p6mftpqFlmPtFAY9N3t-LqQG3mxXnn4uPx4X29KbavT8_rclvUWtuxSLZCX2m0zkViqhR6B9EbJEZNCR1UpGttiTyCCQECM6do0YANxtU4F8v_bvMr9qehOdLwvT-fxR8e60hA</addsrcrecordid><sourcetype>Publisher</sourcetype><iscdi>true</iscdi><recordtype>conference_proceeding</recordtype></control><display><type>conference_proceeding</type><title>Multi-Scale multi-band densenets for audio source separation</title><source>IEEE Xplore All Conference Series</source><creator>Takahashi, Naoya ; Mitsufuji, Yuki</creator><creatorcontrib>Takahashi, Naoya ; Mitsufuji, Yuki</creatorcontrib><description>This paper deals with the problem of audio source separation. To handle the complex and ill-posed nature of the problems of audio source separation, the current state-of-the-art approaches employ deep neural networks to obtain instrumental spectra from a mixture. In this study, we propose a novel network architecture that extends the recently developed densely connected convolutional network (DenseNet), which has shown excellent results on image classification tasks. To deal with the specific problem of audio source separation, an up-sampling layer, block skip connection and band-dedicated dense blocks are incorporated on top of DenseNet. The proposed approach takes advantage of long contextual information and outperforms state-of-the-art results on SiSEC 2016 competition by a large margin in terms of signal-to-distortion ratio. Moreover, the proposed architecture requires significantly fewer parameters and considerably less training time compared with other methods.</description><identifier>EISSN: 1947-1629</identifier><identifier>EISBN: 9781538616321</identifier><identifier>EISBN: 1538616327</identifier><identifier>DOI: 10.1109/WASPAA.2017.8169987</identifier><language>eng</language><publisher>IEEE</publisher><subject>Computer architecture ; Convolution ; convolutional neural networks ; DenseNet ; Kernel ; multi-band ; Source separation ; Spectrogram ; Training</subject><ispartof>2017 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), 2017, p.21-25</ispartof><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c225t-d5b37b235669aeab037619743ae32ad361ba2c25aa73148818eeed953415846c3</citedby></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://ieeexplore.ieee.org/document/8169987$$EHTML$$P50$$Gieee$$H</linktohtml><link.rule.ids>309,310,780,784,789,790,27925,54555,54932</link.rule.ids><linktorsrc>$$Uhttps://ieeexplore.ieee.org/document/8169987$$EView_record_in_IEEE$$FView_record_in_$$GIEEE</linktorsrc></links><search><creatorcontrib>Takahashi, Naoya</creatorcontrib><creatorcontrib>Mitsufuji, Yuki</creatorcontrib><title>Multi-Scale multi-band densenets for audio source separation</title><title>2017 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA)</title><addtitle>WASPAA</addtitle><description>This paper deals with the problem of audio source separation. To handle the complex and ill-posed nature of the problems of audio source separation, the current state-of-the-art approaches employ deep neural networks to obtain instrumental spectra from a mixture. In this study, we propose a novel network architecture that extends the recently developed densely connected convolutional network (DenseNet), which has shown excellent results on image classification tasks. To deal with the specific problem of audio source separation, an up-sampling layer, block skip connection and band-dedicated dense blocks are incorporated on top of DenseNet. The proposed approach takes advantage of long contextual information and outperforms state-of-the-art results on SiSEC 2016 competition by a large margin in terms of signal-to-distortion ratio. Moreover, the proposed architecture requires significantly fewer parameters and considerably less training time compared with other methods.</description><subject>Computer architecture</subject><subject>Convolution</subject><subject>convolutional neural networks</subject><subject>DenseNet</subject><subject>Kernel</subject><subject>multi-band</subject><subject>Source separation</subject><subject>Spectrogram</subject><subject>Training</subject><issn>1947-1629</issn><isbn>9781538616321</isbn><isbn>1538616327</isbn><fulltext>true</fulltext><rsrctype>conference_proceeding</rsrctype><creationdate>2017</creationdate><recordtype>conference_proceeding</recordtype><sourceid>6IE</sourceid><recordid>eNotj81Kw0AURkdBsNY-QTfzAqlz584vuAlFrVBRqOKy3GRuIZImJZMsfHtFu_oOHDjwCbEEtQJQ8e6z3L2V5Uor8KsALsbgL8Qi-gAWgwOHGi7FDKLxBTgdr8VNzl9KWR2Mmon7l6kdm2JXU8vy-McVdUkm7jJ3PGZ56AdJU2p6mftpqFlmPtFAY9N3t-LqQG3mxXnn4uPx4X29KbavT8_rclvUWtuxSLZCX2m0zkViqhR6B9EbJEZNCR1UpGttiTyCCQECM6do0YANxtU4F8v_bvMr9qehOdLwvT-fxR8e60hA</recordid><startdate>201710</startdate><enddate>201710</enddate><creator>Takahashi, Naoya</creator><creator>Mitsufuji, Yuki</creator><general>IEEE</general><scope>6IE</scope><scope>6IL</scope><scope>CBEJK</scope><scope>RIE</scope><scope>RIL</scope></search><sort><creationdate>201710</creationdate><title>Multi-Scale multi-band densenets for audio source separation</title><author>Takahashi, Naoya ; Mitsufuji, Yuki</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c225t-d5b37b235669aeab037619743ae32ad361ba2c25aa73148818eeed953415846c3</frbrgroupid><rsrctype>conference_proceedings</rsrctype><prefilter>conference_proceedings</prefilter><language>eng</language><creationdate>2017</creationdate><topic>Computer architecture</topic><topic>Convolution</topic><topic>convolutional neural networks</topic><topic>DenseNet</topic><topic>Kernel</topic><topic>multi-band</topic><topic>Source separation</topic><topic>Spectrogram</topic><topic>Training</topic><toplevel>online_resources</toplevel><creatorcontrib>Takahashi, Naoya</creatorcontrib><creatorcontrib>Mitsufuji, Yuki</creatorcontrib><collection>IEEE Electronic Library (IEL) Conference Proceedings</collection><collection>IEEE Proceedings Order Plan All Online (POP All Online) 1998-present by volume</collection><collection>IEEE Xplore All Conference Proceedings</collection><collection>IEEE Xplore</collection><collection>IEEE Proceedings Order Plans (POP All) 1998-Present</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Takahashi, Naoya</au><au>Mitsufuji, Yuki</au><format>book</format><genre>proceeding</genre><ristype>CONF</ristype><atitle>Multi-Scale multi-band densenets for audio source separation</atitle><btitle>2017 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA)</btitle><stitle>WASPAA</stitle><date>2017-10</date><risdate>2017</risdate><spage>21</spage><epage>25</epage><pages>21-25</pages><eissn>1947-1629</eissn><eisbn>9781538616321</eisbn><eisbn>1538616327</eisbn><abstract>This paper deals with the problem of audio source separation. To handle the complex and ill-posed nature of the problems of audio source separation, the current state-of-the-art approaches employ deep neural networks to obtain instrumental spectra from a mixture. In this study, we propose a novel network architecture that extends the recently developed densely connected convolutional network (DenseNet), which has shown excellent results on image classification tasks. To deal with the specific problem of audio source separation, an up-sampling layer, block skip connection and band-dedicated dense blocks are incorporated on top of DenseNet. The proposed approach takes advantage of long contextual information and outperforms state-of-the-art results on SiSEC 2016 competition by a large margin in terms of signal-to-distortion ratio. Moreover, the proposed architecture requires significantly fewer parameters and considerably less training time compared with other methods.</abstract><pub>IEEE</pub><doi>10.1109/WASPAA.2017.8169987</doi><tpages>5</tpages></addata></record> |
| fulltext | fulltext_linktorsrc |
| identifier | EISSN: 1947-1629 |
| ispartof | 2017 IEEE Workshop on Applications of Signal Processing to Audio and Acoustics (WASPAA), 2017, p.21-25 |
| issn | 1947-1629 |
| language | eng |
| recordid | cdi_ieee_primary_8169987 |
| source | IEEE Xplore All Conference Series |
| subjects | Computer architecture Convolution convolutional neural networks DenseNet Kernel multi-band Source separation Spectrogram Training |
| title | Multi-Scale multi-band densenets for audio source separation |
| url | http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-26T16%3A17%3A16IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-ieee_CHZPO&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=proceeding&rft.atitle=Multi-Scale%20multi-band%20densenets%20for%20audio%20source%20separation&rft.btitle=2017%20IEEE%20Workshop%20on%20Applications%20of%20Signal%20Processing%20to%20Audio%20and%20Acoustics%20(WASPAA)&rft.au=Takahashi,%20Naoya&rft.date=2017-10&rft.spage=21&rft.epage=25&rft.pages=21-25&rft.eissn=1947-1629&rft_id=info:doi/10.1109/WASPAA.2017.8169987&rft.eisbn=9781538616321&rft.eisbn_list=1538616327&rft_dat=%3Cieee_CHZPO%3E8169987%3C/ieee_CHZPO%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c225t-d5b37b235669aeab037619743ae32ad361ba2c25aa73148818eeed953415846c3%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_id=info:pmid/&rft_ieee_id=8169987&rfr_iscdi=true |