Libri-Light: A Benchmark for ASR with Limited or No Supervision
We introduce a new collection of spoken English audio suitable for training speech recognition systems under limited or no supervision. It is derived from open-source audio books from the LibriVox project. It contains over 60K hours of audio, which is, to our knowledge, the largest freely-available...
Gespeichert in:
| Veröffentlicht in: | arXiv.org 2019-12 |
|---|---|
| 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 | Kahn, Jacob Rivière, Morgane Zheng, Weiyi Kharitonov, Evgeny Xu, Qiantong Pierre-Emmanuel Mazaré Karadayi, Julien Liptchinsky, Vitaliy Collobert, Ronan Fuegen, Christian Likhomanenko, Tatiana Synnaeve, Gabriel Joulin, Armand Abdelrahman, Mohamed Dupoux, Emmanuel |
| description | We introduce a new collection of spoken English audio suitable for training speech recognition systems under limited or no supervision. It is derived from open-source audio books from the LibriVox project. It contains over 60K hours of audio, which is, to our knowledge, the largest freely-available corpus of speech. The audio has been segmented using voice activity detection and is tagged with SNR, speaker ID and genre descriptions. Additionally, we provide baseline systems and evaluation metrics working under three settings: (1) the zero resource/unsupervised setting (ABX), (2) the semi-supervised setting (PER, CER) and (3) the distant supervision setting (WER). Settings (2) and (3) use limited textual resources (10 minutes to 10 hours) aligned with the speech. Setting (3) uses large amounts of unaligned text. They are evaluated on the standard LibriSpeech dev and test sets for comparison with the supervised state-of-the-art. |
| doi_str_mv | 10.48550/arxiv.1912.07875 |
| format | Article |
| fullrecord | <record><control><sourceid>proquest_arxiv</sourceid><recordid>TN_cdi_arxiv_primary_1912_07875</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2328198894</sourcerecordid><originalsourceid>FETCH-LOGICAL-a524-f322db4321ff57f5a68afe12a01ac1451e4c45db787b51df1611f652c85b5b8a3</originalsourceid><addsrcrecordid>eNotj1FLwzAUhYMgOOZ-gE8GfO7MvcltU19kDqdCUXB7L2mb2EzXzrSb-u-tm08HDofD9zF2AWKqNJG4NuHb76eQAk5FohM6YSOUEiKtEM_YpOvWQgiMEySSI3ab-SL4KPNvdX_DZ_zONmW9MeGduzbw2fKVf_m-5pnf-N5WfOieW77cbW3Y-863zTk7deajs5P_HLPV4n41f4yyl4en-SyLDKGKnESsCiURnKPEkYm1cRbQCDAlKAKrSkVVMfAWBJWDGMDFhKWmggpt5JhdHm8Pdvk2-IHxJ_-zzA-Ww-LquNiG9nNnuz5ft7vQDEw5StSQap0q-QthMFMS</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2328198894</pqid></control><display><type>article</type><title>Libri-Light: A Benchmark for ASR with Limited or No Supervision</title><source>arXiv.org</source><source>Free E- Journals</source><creator>Kahn, Jacob ; Rivière, Morgane ; Zheng, Weiyi ; Kharitonov, Evgeny ; Xu, Qiantong ; Pierre-Emmanuel Mazaré ; Karadayi, Julien ; Liptchinsky, Vitaliy ; Collobert, Ronan ; Fuegen, Christian ; Likhomanenko, Tatiana ; Synnaeve, Gabriel ; Joulin, Armand ; Abdelrahman, Mohamed ; Dupoux, Emmanuel</creator><creatorcontrib>Kahn, Jacob ; Rivière, Morgane ; Zheng, Weiyi ; Kharitonov, Evgeny ; Xu, Qiantong ; Pierre-Emmanuel Mazaré ; Karadayi, Julien ; Liptchinsky, Vitaliy ; Collobert, Ronan ; Fuegen, Christian ; Likhomanenko, Tatiana ; Synnaeve, Gabriel ; Joulin, Armand ; Abdelrahman, Mohamed ; Dupoux, Emmanuel</creatorcontrib><description>We introduce a new collection of spoken English audio suitable for training speech recognition systems under limited or no supervision. It is derived from open-source audio books from the LibriVox project. It contains over 60K hours of audio, which is, to our knowledge, the largest freely-available corpus of speech. The audio has been segmented using voice activity detection and is tagged with SNR, speaker ID and genre descriptions. Additionally, we provide baseline systems and evaluation metrics working under three settings: (1) the zero resource/unsupervised setting (ABX), (2) the semi-supervised setting (PER, CER) and (3) the distant supervision setting (WER). Settings (2) and (3) use limited textual resources (10 minutes to 10 hours) aligned with the speech. Setting (3) uses large amounts of unaligned text. They are evaluated on the standard LibriSpeech dev and test sets for comparison with the supervised state-of-the-art.</description><identifier>EISSN: 2331-8422</identifier><identifier>DOI: 10.48550/arxiv.1912.07875</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Computer Science - Computation and Language ; Computer Science - Sound ; Speech recognition ; Supervision ; Systems analysis ; Test sets ; 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/10.1109/ICASSP40776.2020.9052942$$DView published paper (Access to full text may be restricted)$$Hfree_for_read</backlink><backlink>$$Uhttps://doi.org/10.48550/arXiv.1912.07875$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Kahn, Jacob</creatorcontrib><creatorcontrib>Rivière, Morgane</creatorcontrib><creatorcontrib>Zheng, Weiyi</creatorcontrib><creatorcontrib>Kharitonov, Evgeny</creatorcontrib><creatorcontrib>Xu, Qiantong</creatorcontrib><creatorcontrib>Pierre-Emmanuel Mazaré</creatorcontrib><creatorcontrib>Karadayi, Julien</creatorcontrib><creatorcontrib>Liptchinsky, Vitaliy</creatorcontrib><creatorcontrib>Collobert, Ronan</creatorcontrib><creatorcontrib>Fuegen, Christian</creatorcontrib><creatorcontrib>Likhomanenko, Tatiana</creatorcontrib><creatorcontrib>Synnaeve, Gabriel</creatorcontrib><creatorcontrib>Joulin, Armand</creatorcontrib><creatorcontrib>Abdelrahman, Mohamed</creatorcontrib><creatorcontrib>Dupoux, Emmanuel</creatorcontrib><title>Libri-Light: A Benchmark for ASR with Limited or No Supervision</title><title>arXiv.org</title><description>We introduce a new collection of spoken English audio suitable for training speech recognition systems under limited or no supervision. It is derived from open-source audio books from the LibriVox project. It contains over 60K hours of audio, which is, to our knowledge, the largest freely-available corpus of speech. The audio has been segmented using voice activity detection and is tagged with SNR, speaker ID and genre descriptions. Additionally, we provide baseline systems and evaluation metrics working under three settings: (1) the zero resource/unsupervised setting (ABX), (2) the semi-supervised setting (PER, CER) and (3) the distant supervision setting (WER). Settings (2) and (3) use limited textual resources (10 minutes to 10 hours) aligned with the speech. Setting (3) uses large amounts of unaligned text. They are evaluated on the standard LibriSpeech dev and test sets for comparison with the supervised state-of-the-art.</description><subject>Computer Science - Computation and Language</subject><subject>Computer Science - Sound</subject><subject>Speech recognition</subject><subject>Supervision</subject><subject>Systems analysis</subject><subject>Test sets</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>eNotj1FLwzAUhYMgOOZ-gE8GfO7MvcltU19kDqdCUXB7L2mb2EzXzrSb-u-tm08HDofD9zF2AWKqNJG4NuHb76eQAk5FohM6YSOUEiKtEM_YpOvWQgiMEySSI3ab-SL4KPNvdX_DZ_zONmW9MeGduzbw2fKVf_m-5pnf-N5WfOieW77cbW3Y-863zTk7deajs5P_HLPV4n41f4yyl4en-SyLDKGKnESsCiURnKPEkYm1cRbQCDAlKAKrSkVVMfAWBJWDGMDFhKWmggpt5JhdHm8Pdvk2-IHxJ_-zzA-Ww-LquNiG9nNnuz5ft7vQDEw5StSQap0q-QthMFMS</recordid><startdate>20191217</startdate><enddate>20191217</enddate><creator>Kahn, Jacob</creator><creator>Rivière, Morgane</creator><creator>Zheng, Weiyi</creator><creator>Kharitonov, Evgeny</creator><creator>Xu, Qiantong</creator><creator>Pierre-Emmanuel Mazaré</creator><creator>Karadayi, Julien</creator><creator>Liptchinsky, Vitaliy</creator><creator>Collobert, Ronan</creator><creator>Fuegen, Christian</creator><creator>Likhomanenko, Tatiana</creator><creator>Synnaeve, Gabriel</creator><creator>Joulin, Armand</creator><creator>Abdelrahman, Mohamed</creator><creator>Dupoux, Emmanuel</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>20191217</creationdate><title>Libri-Light: A Benchmark for ASR with Limited or No Supervision</title><author>Kahn, Jacob ; Rivière, Morgane ; Zheng, Weiyi ; Kharitonov, Evgeny ; Xu, Qiantong ; Pierre-Emmanuel Mazaré ; Karadayi, Julien ; Liptchinsky, Vitaliy ; Collobert, Ronan ; Fuegen, Christian ; Likhomanenko, Tatiana ; Synnaeve, Gabriel ; Joulin, Armand ; Abdelrahman, Mohamed ; Dupoux, Emmanuel</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a524-f322db4321ff57f5a68afe12a01ac1451e4c45db787b51df1611f652c85b5b8a3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2019</creationdate><topic>Computer Science - Computation and Language</topic><topic>Computer Science - Sound</topic><topic>Speech recognition</topic><topic>Supervision</topic><topic>Systems analysis</topic><topic>Test sets</topic><topic>Voice activity detectors</topic><topic>Voice recognition</topic><toplevel>online_resources</toplevel><creatorcontrib>Kahn, Jacob</creatorcontrib><creatorcontrib>Rivière, Morgane</creatorcontrib><creatorcontrib>Zheng, Weiyi</creatorcontrib><creatorcontrib>Kharitonov, Evgeny</creatorcontrib><creatorcontrib>Xu, Qiantong</creatorcontrib><creatorcontrib>Pierre-Emmanuel Mazaré</creatorcontrib><creatorcontrib>Karadayi, Julien</creatorcontrib><creatorcontrib>Liptchinsky, Vitaliy</creatorcontrib><creatorcontrib>Collobert, Ronan</creatorcontrib><creatorcontrib>Fuegen, Christian</creatorcontrib><creatorcontrib>Likhomanenko, Tatiana</creatorcontrib><creatorcontrib>Synnaeve, Gabriel</creatorcontrib><creatorcontrib>Joulin, Armand</creatorcontrib><creatorcontrib>Abdelrahman, Mohamed</creatorcontrib><creatorcontrib>Dupoux, Emmanuel</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>Access via ProQuest (Open Access)</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>Kahn, Jacob</au><au>Rivière, Morgane</au><au>Zheng, Weiyi</au><au>Kharitonov, Evgeny</au><au>Xu, Qiantong</au><au>Pierre-Emmanuel Mazaré</au><au>Karadayi, Julien</au><au>Liptchinsky, Vitaliy</au><au>Collobert, Ronan</au><au>Fuegen, Christian</au><au>Likhomanenko, Tatiana</au><au>Synnaeve, Gabriel</au><au>Joulin, Armand</au><au>Abdelrahman, Mohamed</au><au>Dupoux, Emmanuel</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Libri-Light: A Benchmark for ASR with Limited or No Supervision</atitle><jtitle>arXiv.org</jtitle><date>2019-12-17</date><risdate>2019</risdate><eissn>2331-8422</eissn><abstract>We introduce a new collection of spoken English audio suitable for training speech recognition systems under limited or no supervision. It is derived from open-source audio books from the LibriVox project. It contains over 60K hours of audio, which is, to our knowledge, the largest freely-available corpus of speech. The audio has been segmented using voice activity detection and is tagged with SNR, speaker ID and genre descriptions. Additionally, we provide baseline systems and evaluation metrics working under three settings: (1) the zero resource/unsupervised setting (ABX), (2) the semi-supervised setting (PER, CER) and (3) the distant supervision setting (WER). Settings (2) and (3) use limited textual resources (10 minutes to 10 hours) aligned with the speech. Setting (3) uses large amounts of unaligned text. They are evaluated on the standard LibriSpeech dev and test sets for comparison with the supervised state-of-the-art.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><doi>10.48550/arxiv.1912.07875</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_1912_07875 |
| source | arXiv.org; Free E- Journals |
| subjects | Computer Science - Computation and Language Computer Science - Sound Speech recognition Supervision Systems analysis Test sets Voice activity detectors Voice recognition |
| title | Libri-Light: A Benchmark for ASR with Limited or No Supervision |
| url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-20T07%3A23%3A42IST&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=Libri-Light:%20A%20Benchmark%20for%20ASR%20with%20Limited%20or%20No%20Supervision&rft.jtitle=arXiv.org&rft.au=Kahn,%20Jacob&rft.date=2019-12-17&rft.eissn=2331-8422&rft_id=info:doi/10.48550/arxiv.1912.07875&rft_dat=%3Cproquest_arxiv%3E2328198894%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=2328198894&rft_id=info:pmid/&rfr_iscdi=true |