Is Someone Speaking? Exploring Long-term Temporal Features for Audio-visual Active Speaker Detection
Active speaker detection (ASD) seeks to detect who is speaking in a visual scene of one or more speakers. The successful ASD depends on accurate interpretation of short-term and long-term audio and visual information, as well as audio-visual interaction. Unlike the prior work where systems make deci...
Gespeichert in:
| Veröffentlicht in: | arXiv.org 2021-07 |
|---|---|
| Hauptverfasser: | , , , , , |
| 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 | Tao, Ruijie Pan, Zexu Das, Rohan Kumar Qian, Xinyuan Mike Zheng Shou Li, Haizhou |
| description | Active speaker detection (ASD) seeks to detect who is speaking in a visual scene of one or more speakers. The successful ASD depends on accurate interpretation of short-term and long-term audio and visual information, as well as audio-visual interaction. Unlike the prior work where systems make decision instantaneously using short-term features, we propose a novel framework, named TalkNet, that makes decision by taking both short-term and long-term features into consideration. TalkNet consists of audio and visual temporal encoders for feature representation, audio-visual cross-attention mechanism for inter-modality interaction, and a self-attention mechanism to capture long-term speaking evidence. The experiments demonstrate that TalkNet achieves 3.5% and 2.2% improvement over the state-of-the-art systems on the AVA-ActiveSpeaker dataset and Columbia ASD dataset, respectively. Code has been made available at: https://github.com/TaoRuijie/TalkNet_ASD. |
| doi_str_mv | 10.48550/arxiv.2107.06592 |
| format | Article |
| fullrecord | <record><control><sourceid>proquest_arxiv</sourceid><recordid>TN_cdi_arxiv_primary_2107_06592</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2551800769</sourcerecordid><originalsourceid>FETCH-LOGICAL-a529-cb4cdcafe8dfbfa0826d1001f1c26184a3c8ccd21612399928063077a19bd0e3</originalsourceid><addsrcrecordid>eNotkFFrwjAUhcNgMHH-gD0tsOe6m6RNk6chTjdB2IO-l5jcSp1tuqQV9-_XqU_3cM_hcPgIeWIwTVWWwasJ5-o05QzyKchM8zsy4kKwRKWcP5BJjAcA4DLnWSZGxK0i3fgafYN006L5rpr9G12c26MPg6Rr3-yTDkNNt1i3PpgjXaLp-oCRlj7QWe8qn5yq2A_OzHbV6daDgb5jh8PHN4_kvjTHiJPbHZPNcrGdfybrr4_VfLZOTMZ1YnepddaUqFy5Kw0oLh0DYCWzXDKVGmGVtY4zybjQWnMFUkCeG6Z3DlCMyfO19UKgaENVm_Bb_JMoLiSGxMs10Qb_02PsioPvQzNMKgYaTAHkUos_e6FhiQ</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2551800769</pqid></control><display><type>article</type><title>Is Someone Speaking? Exploring Long-term Temporal Features for Audio-visual Active Speaker Detection</title><source>Freely Accessible Journals</source><source>arXiv.org</source><creator>Tao, Ruijie ; Pan, Zexu ; Das, Rohan Kumar ; Qian, Xinyuan ; Mike Zheng Shou ; Li, Haizhou</creator><creatorcontrib>Tao, Ruijie ; Pan, Zexu ; Das, Rohan Kumar ; Qian, Xinyuan ; Mike Zheng Shou ; Li, Haizhou</creatorcontrib><description>Active speaker detection (ASD) seeks to detect who is speaking in a visual scene of one or more speakers. The successful ASD depends on accurate interpretation of short-term and long-term audio and visual information, as well as audio-visual interaction. Unlike the prior work where systems make decision instantaneously using short-term features, we propose a novel framework, named TalkNet, that makes decision by taking both short-term and long-term features into consideration. TalkNet consists of audio and visual temporal encoders for feature representation, audio-visual cross-attention mechanism for inter-modality interaction, and a self-attention mechanism to capture long-term speaking evidence. The experiments demonstrate that TalkNet achieves 3.5% and 2.2% improvement over the state-of-the-art systems on the AVA-ActiveSpeaker dataset and Columbia ASD dataset, respectively. Code has been made available at: https://github.com/TaoRuijie/TalkNet_ASD.</description><identifier>EISSN: 2331-8422</identifier><identifier>DOI: 10.48550/arxiv.2107.06592</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Audio data ; Audio equipment ; Coders ; Computer Science - Sound ; Datasets</subject><ispartof>arXiv.org, 2021-07</ispartof><rights>2021. 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>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,784,885,27923</link.rule.ids><backlink>$$Uhttps://doi.org/10.48550/arXiv.2107.06592$$DView paper in arXiv$$Hfree_for_read</backlink><backlink>$$Uhttps://doi.org/10.1145/3474085.3475587$$DView published paper (Access to full text may be restricted)$$Hfree_for_read</backlink></links><search><creatorcontrib>Tao, Ruijie</creatorcontrib><creatorcontrib>Pan, Zexu</creatorcontrib><creatorcontrib>Das, Rohan Kumar</creatorcontrib><creatorcontrib>Qian, Xinyuan</creatorcontrib><creatorcontrib>Mike Zheng Shou</creatorcontrib><creatorcontrib>Li, Haizhou</creatorcontrib><title>Is Someone Speaking? Exploring Long-term Temporal Features for Audio-visual Active Speaker Detection</title><title>arXiv.org</title><description>Active speaker detection (ASD) seeks to detect who is speaking in a visual scene of one or more speakers. The successful ASD depends on accurate interpretation of short-term and long-term audio and visual information, as well as audio-visual interaction. Unlike the prior work where systems make decision instantaneously using short-term features, we propose a novel framework, named TalkNet, that makes decision by taking both short-term and long-term features into consideration. TalkNet consists of audio and visual temporal encoders for feature representation, audio-visual cross-attention mechanism for inter-modality interaction, and a self-attention mechanism to capture long-term speaking evidence. The experiments demonstrate that TalkNet achieves 3.5% and 2.2% improvement over the state-of-the-art systems on the AVA-ActiveSpeaker dataset and Columbia ASD dataset, respectively. Code has been made available at: https://github.com/TaoRuijie/TalkNet_ASD.</description><subject>Audio data</subject><subject>Audio equipment</subject><subject>Coders</subject><subject>Computer Science - Sound</subject><subject>Datasets</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</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>eNotkFFrwjAUhcNgMHH-gD0tsOe6m6RNk6chTjdB2IO-l5jcSp1tuqQV9-_XqU_3cM_hcPgIeWIwTVWWwasJ5-o05QzyKchM8zsy4kKwRKWcP5BJjAcA4DLnWSZGxK0i3fgafYN006L5rpr9G12c26MPg6Rr3-yTDkNNt1i3PpgjXaLp-oCRlj7QWe8qn5yq2A_OzHbV6daDgb5jh8PHN4_kvjTHiJPbHZPNcrGdfybrr4_VfLZOTMZ1YnepddaUqFy5Kw0oLh0DYCWzXDKVGmGVtY4zybjQWnMFUkCeG6Z3DlCMyfO19UKgaENVm_Bb_JMoLiSGxMs10Qb_02PsioPvQzNMKgYaTAHkUos_e6FhiQ</recordid><startdate>20210725</startdate><enddate>20210725</enddate><creator>Tao, Ruijie</creator><creator>Pan, Zexu</creator><creator>Das, Rohan Kumar</creator><creator>Qian, Xinyuan</creator><creator>Mike Zheng Shou</creator><creator>Li, Haizhou</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>20210725</creationdate><title>Is Someone Speaking? Exploring Long-term Temporal Features for Audio-visual Active Speaker Detection</title><author>Tao, Ruijie ; Pan, Zexu ; Das, Rohan Kumar ; Qian, Xinyuan ; Mike Zheng Shou ; Li, Haizhou</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a529-cb4cdcafe8dfbfa0826d1001f1c26184a3c8ccd21612399928063077a19bd0e3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><topic>Audio data</topic><topic>Audio equipment</topic><topic>Coders</topic><topic>Computer Science - Sound</topic><topic>Datasets</topic><toplevel>online_resources</toplevel><creatorcontrib>Tao, Ruijie</creatorcontrib><creatorcontrib>Pan, Zexu</creatorcontrib><creatorcontrib>Das, Rohan Kumar</creatorcontrib><creatorcontrib>Qian, Xinyuan</creatorcontrib><creatorcontrib>Mike Zheng Shou</creatorcontrib><creatorcontrib>Li, Haizhou</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science & 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>Tao, Ruijie</au><au>Pan, Zexu</au><au>Das, Rohan Kumar</au><au>Qian, Xinyuan</au><au>Mike Zheng Shou</au><au>Li, Haizhou</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Is Someone Speaking? Exploring Long-term Temporal Features for Audio-visual Active Speaker Detection</atitle><jtitle>arXiv.org</jtitle><date>2021-07-25</date><risdate>2021</risdate><eissn>2331-8422</eissn><abstract>Active speaker detection (ASD) seeks to detect who is speaking in a visual scene of one or more speakers. The successful ASD depends on accurate interpretation of short-term and long-term audio and visual information, as well as audio-visual interaction. Unlike the prior work where systems make decision instantaneously using short-term features, we propose a novel framework, named TalkNet, that makes decision by taking both short-term and long-term features into consideration. TalkNet consists of audio and visual temporal encoders for feature representation, audio-visual cross-attention mechanism for inter-modality interaction, and a self-attention mechanism to capture long-term speaking evidence. The experiments demonstrate that TalkNet achieves 3.5% and 2.2% improvement over the state-of-the-art systems on the AVA-ActiveSpeaker dataset and Columbia ASD dataset, respectively. Code has been made available at: https://github.com/TaoRuijie/TalkNet_ASD.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><doi>10.48550/arxiv.2107.06592</doi><oa>free_for_read</oa></addata></record> |
| fulltext | fulltext |
| identifier | EISSN: 2331-8422 |
| ispartof | arXiv.org, 2021-07 |
| issn | 2331-8422 |
| language | eng |
| recordid | cdi_arxiv_primary_2107_06592 |
| source | Freely Accessible Journals; arXiv.org |
| subjects | Audio data Audio equipment Coders Computer Science - Sound Datasets |
| title | Is Someone Speaking? Exploring Long-term Temporal Features for Audio-visual Active Speaker Detection |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-13T19%3A29%3A14IST&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=Is%20Someone%20Speaking?%20Exploring%20Long-term%20Temporal%20Features%20for%20Audio-visual%20Active%20Speaker%20Detection&rft.jtitle=arXiv.org&rft.au=Tao,%20Ruijie&rft.date=2021-07-25&rft.eissn=2331-8422&rft_id=info:doi/10.48550/arxiv.2107.06592&rft_dat=%3Cproquest_arxiv%3E2551800769%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=2551800769&rft_id=info:pmid/&rfr_iscdi=true |