{
  "id": 311245,
  "title": "VERY Long (1.25 Hours) Training Audio Clip - 'commyn/XC548866.ogg'",
  "url": "/competitions/birdclef-2022/discussion/311245",
  "author_name": "",
  "post_date": "2022-03-05T18:23:30.502779300Z",
  "votes": 23,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I guess I'm just pointing this out. It will crash your notebook for instance if you try to plot the whole spectrogram.</p>\n<p>Here's a screenshot showing the clip and the length:</p>\n<p><img src=\"https://i.ibb.co/d0y5zYs/Screen-Shot-2022-03-05-at-1-00-10-PM.png\" alt=\"long audio\"></p>\n<p>Just for reference, here is a histogram of clip durations in the training data showing that this is mostly an anomalous length. Note that the y-axis is logged to allow for better viewing:</p>\n<p><img src=\"https://i.ibb.co/KDmfJC3/Screen-Shot-2022-03-05-at-1-12-15-PM.png\" alt=\"audio histogram\"></p>\n<p>As you can see there are only a few clips that are really long… if you really want to know. Here's a printout showing the various lengths bucketed in my random way. You'll also notice some REALLLLLY short clips too! Shorter than 0.1 seconds (0.03484375 seconds to be precise)!</p>\n<p><img src=\"https://i.ibb.co/gvY81Gn/Screen-Shot-2022-03-05-at-1-26-25-PM.png\" alt=\"randombuckets\"></p>\n<p>This is my first time with an audio competition so maybe this is normal. I haven't examined the previous competition's data either… so I'm not sure if we had things like this previously.</p>",
  "messages": [
    {
      "id": "1713190",
      "postDate": "03/05/2022 18:23:30",
      "content": "<p>I guess I'm just pointing this out. It will crash your notebook for instance if you try to plot the whole spectrogram.</p>\n<p>Here's a screenshot showing the clip and the length:</p>\n<p><img src=\"https://i.ibb.co/d0y5zYs/Screen-Shot-2022-03-05-at-1-00-10-PM.png\" alt=\"long audio\"></p>\n<p>Just for reference, here is a histogram of clip durations in the training data showing that this is mostly an anomalous length. Note that the y-axis is logged to allow for better viewing:</p>\n<p><img src=\"https://i.ibb.co/KDmfJC3/Screen-Shot-2022-03-05-at-1-12-15-PM.png\" alt=\"audio histogram\"></p>\n<p>As you can see there are only a few clips that are really long… if you really want to know. Here's a printout showing the various lengths bucketed in my random way. You'll also notice some REALLLLLY short clips too! Shorter than 0.1 seconds (0.03484375 seconds to be precise)!</p>\n<p><img src=\"https://i.ibb.co/gvY81Gn/Screen-Shot-2022-03-05-at-1-26-25-PM.png\" alt=\"randombuckets\"></p>\n<p>This is my first time with an audio competition so maybe this is normal. I haven't examined the previous competition's data either… so I'm not sure if we had things like this previously.</p>",
      "rawMarkdown": "I guess I'm just pointing this out. It will crash your notebook for instance if you try to plot the whole spectrogram.\n\nHere's a screenshot showing the clip and the length:\n\n![long audio](https://i.ibb.co/d0y5zYs/Screen-Shot-2022-03-05-at-1-00-10-PM.png)\n\nJust for reference, here is a histogram of clip durations in the training data showing that this is mostly an anomalous length. Note that the y-axis is logged to allow for better viewing:\n\n![audio histogram](https://i.ibb.co/KDmfJC3/Screen-Shot-2022-03-05-at-1-12-15-PM.png)\n\nAs you can see there are only a few clips that are really long... if you really want to know. Here's a printout showing the various lengths bucketed in my random way. You'll also notice some REALLLLLY short clips too! Shorter than 0.1 seconds (0.03484375 seconds to be precise)!\n\n![randombuckets](https://i.ibb.co/gvY81Gn/Screen-Shot-2022-03-05-at-1-26-25-PM.png)\n\nThis is my first time with an audio competition so maybe this is normal. I haven't examined the previous competition's data either... so I'm not sure if we had things like this previously.",
      "votes": null
    },
    {
      "id": "1713511",
      "postDate": "03/06/2022 05:25:01",
      "content": "<p>Thank you for sharing👍</p>",
      "rawMarkdown": "Thank you for sharing👍",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1713511,
      "author_name": "imnoob",
      "author_url": "",
      "post_date": "03/06/2022 05:25:01",
      "content": "<p>Thank you for sharing👍</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1713190": "I guess I'm just pointing this out. It will crash your notebook for instance if you try to plot the whole spectrogram.\n\nHere's a screenshot showing the clip and the length:\n\n![long audio](https://i.ibb.co/d0y5zYs/Screen-Shot-2022-03-05-at-1-00-10-PM.png)\n\nJust for reference, here is a histogram of clip durations in the training data showing that this is mostly an anomalous length. Note that the y-axis is logged to allow for better viewing:\n\n![audio histogram](https://i.ibb.co/KDmfJC3/Screen-Shot-2022-03-05-at-1-12-15-PM.png)\n\nAs you can see there are only a few clips that are really long... if you really want to know. Here's a printout showing the various lengths bucketed in my random way. You'll also notice some REALLLLLY short clips too! Shorter than 0.1 seconds (0.03484375 seconds to be precise)!\n\n![randombuckets](https://i.ibb.co/gvY81Gn/Screen-Shot-2022-03-05-at-1-26-25-PM.png)\n\nThis is my first time with an audio competition so maybe this is normal. I haven't examined the previous competition's data either... so I'm not sure if we had things like this previously.",
    "1713511": "Thank you for sharing👍"
  },
  "source": "meta"
}