{
  "id": 172693,
  "title": "Statistical Tests as Features",
  "url": "/competitions/birdsong-recognition/discussion/172693",
  "author_name": "",
  "post_date": "2020-08-06T05:38:46.898570700Z",
  "votes": 3,
  "comment_count": 1,
  "views": 0,
  "content": "<p>This is probably a dumb question, but for those segmenting their audio data to have a good majority of the birdcall dominate the recording, has anyone tried using the results of statistical tests like kstest and so forth to generate features (such as using the pvalue and statistic returned from a kstest of every recording compared to a general batch of audio belonging to each bird class). </p>",
  "messages": [
    {
      "id": "960061",
      "postDate": "08/06/2020 05:38:46",
      "content": "<p>This is probably a dumb question, but for those segmenting their audio data to have a good majority of the birdcall dominate the recording, has anyone tried using the results of statistical tests like kstest and so forth to generate features (such as using the pvalue and statistic returned from a kstest of every recording compared to a general batch of audio belonging to each bird class). </p>",
      "rawMarkdown": "This is probably a dumb question, but for those segmenting their audio data to have a good majority of the birdcall dominate the recording, has anyone tried using the results of statistical tests like kstest and so forth to generate features (such as using the pvalue and statistic returned from a kstest of every recording compared to a general batch of audio belonging to each bird class).",
      "votes": null
    },
    {
      "id": "960067",
      "postDate": "08/06/2020 05:47:52",
      "content": "<p>I also have interest in the use of fractal dimensions and entropy as features</p>",
      "rawMarkdown": "I also have interest in the use of fractal dimensions and entropy as features",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 960067,
      "author_name": "eladwar",
      "author_url": "",
      "post_date": "08/06/2020 05:47:52",
      "content": "<p>I also have interest in the use of fractal dimensions and entropy as features</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "960061": "This is probably a dumb question, but for those segmenting their audio data to have a good majority of the birdcall dominate the recording, has anyone tried using the results of statistical tests like kstest and so forth to generate features (such as using the pvalue and statistic returned from a kstest of every recording compared to a general batch of audio belonging to each bird class).",
    "960067": "I also have interest in the use of fractal dimensions and entropy as features"
  },
  "source": "meta"
}