{
  "id": 127658,
  "title": "Releasing test set of FSDKaggle2019 dataset used in this competition",
  "url": "/competitions/freesound-audio-tagging-2019/discussion/127658",
  "author_name": "",
  "post_date": "2020-01-25T14:03:18.749149900Z",
  "votes": 9,
  "comment_count": 4,
  "views": 0,
  "content": "<p>Dear all,</p>\n\n<p>We’re glad to announce we have released the full test set &amp; labels of the FSDKaggle2019 dataset used in this competition.</p>\n\n<p><strong>FSDKaggle2019 is freely available from Zenodo (including the full test set and labels as well as updated csv files with metadata and licenses)</strong>: <strong><a href=\"https://doi.org/10.5281/zenodo.3612637\">https://doi.org/10.5281/zenodo.3612637</a></strong></p>\n\n<p>You can find more details in our DCASE 2019 paper:</p>\n\n<blockquote>\n  <p>Eduardo Fonseca, Manoj Plakal, Frederic Font, Daniel P. W. Ellis, Xavier Serra. <em>Audio tagging with noisy labels and minimal supervision</em>. In Proceedings of DCASE2019 Workshop, NYC, US (2019). URL: <a href=\"https://arxiv.org/abs/1906.02975\">https://arxiv.org/abs/1906.02975</a></p>\n</blockquote>\n\n<p>FSDKaggle2019 includes almost 30k audio clips amounting over 100h of audio, encompassing 80 classes drawn from the AudioSet Ontology. It includes a human curated train set from Freesound (~5k clips, ~11h), a noisy train set from Flickr (~20k clips, ~80h), and a test set from Freesound (~4.5k clips, ~13h). The dataset allows development and evaluation of machine listening methods in conditions of label noise, minimal supervision, and real-world acoustic mismatch.</p>\n\n<p>Both competition and dataset have been a collaboration between the Music Technology Group of Universitat Pompeu Fabra, and the Sound Understanding team at Google AI Perception. This effort was kindly sponsored by a Google Faculty Research Award 2018.</p>\n\n<p>Best,</p>\n\n<p>Eduardo, Manoj, Frederic, Dan and Xavier</p>",
  "messages": [
    {
      "id": "728958",
      "postDate": "01/25/2020 14:03:18",
      "content": "<p>Dear all,</p>\n\n<p>We’re glad to announce we have released the full test set &amp; labels of the FSDKaggle2019 dataset used in this competition.</p>\n\n<p><strong>FSDKaggle2019 is freely available from Zenodo (including the full test set and labels as well as updated csv files with metadata and licenses)</strong>: <strong><a href=\"https://doi.org/10.5281/zenodo.3612637\">https://doi.org/10.5281/zenodo.3612637</a></strong></p>\n\n<p>You can find more details in our DCASE 2019 paper:</p>\n\n<blockquote>\n  <p>Eduardo Fonseca, Manoj Plakal, Frederic Font, Daniel P. W. Ellis, Xavier Serra. <em>Audio tagging with noisy labels and minimal supervision</em>. In Proceedings of DCASE2019 Workshop, NYC, US (2019). URL: <a href=\"https://arxiv.org/abs/1906.02975\">https://arxiv.org/abs/1906.02975</a></p>\n</blockquote>\n\n<p>FSDKaggle2019 includes almost 30k audio clips amounting over 100h of audio, encompassing 80 classes drawn from the AudioSet Ontology. It includes a human curated train set from Freesound (~5k clips, ~11h), a noisy train set from Flickr (~20k clips, ~80h), and a test set from Freesound (~4.5k clips, ~13h). The dataset allows development and evaluation of machine listening methods in conditions of label noise, minimal supervision, and real-world acoustic mismatch.</p>\n\n<p>Both competition and dataset have been a collaboration between the Music Technology Group of Universitat Pompeu Fabra, and the Sound Understanding team at Google AI Perception. This effort was kindly sponsored by a Google Faculty Research Award 2018.</p>\n\n<p>Best,</p>\n\n<p>Eduardo, Manoj, Frederic, Dan and Xavier</p>",
      "rawMarkdown": "Dear all,\n\nWe’re glad to announce we have released the full test set &amp; labels of the FSDKaggle2019 dataset used in this competition.\n\n**FSDKaggle2019 is freely available from Zenodo (including the full test set and labels as well as updated csv files with metadata and licenses)**: <a href=\"https://doi.org/10.5281/zenodo.3612637\" target=\"_blank\">**https://doi.org/10.5281/zenodo.3612637**</a>\n\nYou can find more details in our DCASE 2019 paper:\n&gt; Eduardo Fonseca, Manoj Plakal, Frederic Font, Daniel P. W. Ellis, Xavier Serra. *Audio tagging with noisy labels and minimal supervision*. In Proceedings of DCASE2019 Workshop, NYC, US (2019). URL: https://arxiv.org/abs/1906.02975\n\nFSDKaggle2019 includes almost 30k audio clips amounting over 100h of audio, encompassing 80 classes drawn from the AudioSet Ontology. It includes a human curated train set from Freesound (~5k clips, ~11h), a noisy train set from Flickr (~20k clips, ~80h), and a test set from Freesound (~4.5k clips, ~13h). The dataset allows development and evaluation of machine listening methods in conditions of label noise, minimal supervision, and real-world acoustic mismatch.\n\nBoth competition and dataset have been a collaboration between the Music Technology Group of Universitat Pompeu Fabra, and the Sound Understanding team at Google AI Perception. This effort was kindly sponsored by a Google Faculty Research Award 2018.\n\nBest,\n\nEduardo, Manoj, Frederic, Dan and Xavier",
      "votes": null
    },
    {
      "id": "739714",
      "postDate": "02/08/2020 08:41:21",
      "content": "<p>Nice</p>",
      "rawMarkdown": "Nice",
      "votes": null
    },
    {
      "id": "774929",
      "postDate": "03/16/2020 04:17:15",
      "content": "<p>Will there be a freesound competition in 2020?</p>",
      "rawMarkdown": "Will there be a freesound competition in 2020?",
      "votes": null
    },
    {
      "id": "795691",
      "postDate": "04/03/2020 00:37:36",
      "content": "<p>Will there be a freesound comp coming soon?</p>",
      "rawMarkdown": "Will there be a freesound comp coming soon?",
      "votes": null
    },
    {
      "id": "796354",
      "postDate": "04/03/2020 14:21:32",
      "content": "<p>Hi all, </p>\n\n<p>Thanks for the interest! We don't plan to host a Freesound competition this year. We'll let you know if we plan to do it for coming years.</p>\n\n<p>However, the <strong>Detection and Classification of Acoustic Scenes and Events (DCASE) Challenge 2020</strong> has recently launched, including several interesting tasks:</p>\n\n<p>Task 1, Acoustic Scene Classification\nTask 2, Unsupervised Detection of Anomalous Sounds for Machine Condition Monitoring <br>\nTask 3, Sound Event Localization and Detection <br>\nTask 4, Sound Event Detection and Separation in Domestic Environments <br>\nTask 5, Urban Sound Tagging with Spatiotemporal Context \nTask 6, Automated Audio Captioning</p>\n\n<p>More info at:\n<a href=\"http://dcase.community/challenge2020/index\">http://dcase.community/challenge2020/index</a></p>\n\n<p>Also, let us know if you have any DCASE related questions!</p>\n\n<p>Eduardo, on behalf of the organizers.</p>",
      "rawMarkdown": "Hi all, \n\nThanks for the interest! We don't plan to host a Freesound competition this year. We'll let you know if we plan to do it for coming years.\n\nHowever, the **Detection and Classification of Acoustic Scenes and Events (DCASE) Challenge 2020** has recently launched, including several interesting tasks:\n\nTask 1, Acoustic Scene Classification\nTask 2, Unsupervised Detection of Anomalous Sounds for Machine Condition Monitoring \t\nTask 3, Sound Event Localization and Detection \t\nTask 4, Sound Event Detection and Separation in Domestic Environments \t\nTask 5, Urban Sound Tagging with Spatiotemporal Context \nTask 6, Automated Audio Captioning\n\nMore info at:\nhttp://dcase.community/challenge2020/index\n\nAlso, let us know if you have any DCASE related questions!\n\nEduardo, on behalf of the organizers.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 739714,
      "author_name": "mishraanup1982",
      "author_url": "",
      "post_date": "02/08/2020 08:41:21",
      "content": "<p>Nice</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 774929,
      "author_name": "pylsit",
      "author_url": "",
      "post_date": "03/16/2020 04:17:15",
      "content": "<p>Will there be a freesound competition in 2020?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 795691,
      "author_name": "tanlikesmath",
      "author_url": "",
      "post_date": "04/03/2020 00:37:36",
      "content": "<p>Will there be a freesound comp coming soon?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 796354,
      "author_name": "eduardofonseca",
      "author_url": "",
      "post_date": "04/03/2020 14:21:32",
      "content": "<p>Hi all, </p>\n\n<p>Thanks for the interest! We don't plan to host a Freesound competition this year. We'll let you know if we plan to do it for coming years.</p>\n\n<p>However, the <strong>Detection and Classification of Acoustic Scenes and Events (DCASE) Challenge 2020</strong> has recently launched, including several interesting tasks:</p>\n\n<p>Task 1, Acoustic Scene Classification\nTask 2, Unsupervised Detection of Anomalous Sounds for Machine Condition Monitoring <br>\nTask 3, Sound Event Localization and Detection <br>\nTask 4, Sound Event Detection and Separation in Domestic Environments <br>\nTask 5, Urban Sound Tagging with Spatiotemporal Context \nTask 6, Automated Audio Captioning</p>\n\n<p>More info at:\n<a href=\"http://dcase.community/challenge2020/index\">http://dcase.community/challenge2020/index</a></p>\n\n<p>Also, let us know if you have any DCASE related questions!</p>\n\n<p>Eduardo, on behalf of the organizers.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "728958": "Dear all,\n\nWe’re glad to announce we have released the full test set &amp; labels of the FSDKaggle2019 dataset used in this competition.\n\n**FSDKaggle2019 is freely available from Zenodo (including the full test set and labels as well as updated csv files with metadata and licenses)**: <a href=\"https://doi.org/10.5281/zenodo.3612637\" target=\"_blank\">**https://doi.org/10.5281/zenodo.3612637**</a>\n\nYou can find more details in our DCASE 2019 paper:\n&gt; Eduardo Fonseca, Manoj Plakal, Frederic Font, Daniel P. W. Ellis, Xavier Serra. *Audio tagging with noisy labels and minimal supervision*. In Proceedings of DCASE2019 Workshop, NYC, US (2019). URL: https://arxiv.org/abs/1906.02975\n\nFSDKaggle2019 includes almost 30k audio clips amounting over 100h of audio, encompassing 80 classes drawn from the AudioSet Ontology. It includes a human curated train set from Freesound (~5k clips, ~11h), a noisy train set from Flickr (~20k clips, ~80h), and a test set from Freesound (~4.5k clips, ~13h). The dataset allows development and evaluation of machine listening methods in conditions of label noise, minimal supervision, and real-world acoustic mismatch.\n\nBoth competition and dataset have been a collaboration between the Music Technology Group of Universitat Pompeu Fabra, and the Sound Understanding team at Google AI Perception. This effort was kindly sponsored by a Google Faculty Research Award 2018.\n\nBest,\n\nEduardo, Manoj, Frederic, Dan and Xavier",
    "739714": "Nice",
    "774929": "Will there be a freesound competition in 2020?",
    "795691": "Will there be a freesound comp coming soon?",
    "796354": "Hi all, \n\nThanks for the interest! We don't plan to host a Freesound competition this year. We'll let you know if we plan to do it for coming years.\n\nHowever, the **Detection and Classification of Acoustic Scenes and Events (DCASE) Challenge 2020** has recently launched, including several interesting tasks:\n\nTask 1, Acoustic Scene Classification\nTask 2, Unsupervised Detection of Anomalous Sounds for Machine Condition Monitoring \t\nTask 3, Sound Event Localization and Detection \t\nTask 4, Sound Event Detection and Separation in Domestic Environments \t\nTask 5, Urban Sound Tagging with Spatiotemporal Context \nTask 6, Automated Audio Captioning\n\nMore info at:\nhttp://dcase.community/challenge2020/index\n\nAlso, let us know if you have any DCASE related questions!\n\nEduardo, on behalf of the organizers."
  },
  "source": "meta"
}