{
  "id": 233301,
  "title": "How to perform inference?",
  "url": "/competitions/birdclef-2021/discussion/233301",
  "author_name": "",
  "post_date": "2021-04-18T13:30:32.405586200Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I am new to audio competitions. Sorry if this might seem like a silly / easy question.</p>\n<p>We are given the labels for each 5-second window of each audio file. For inference, how do we process the 10-second audio files in such a way that we can make inference? Do we segment the test audio files into windows and average the predictions on each window? Not to mention there is a possibility that there is no bird in each audio file.</p>",
  "messages": [
    {
      "id": "1277175",
      "postDate": "04/18/2021 13:30:32",
      "content": "<p>I am new to audio competitions. Sorry if this might seem like a silly / easy question.</p>\n<p>We are given the labels for each 5-second window of each audio file. For inference, how do we process the 10-second audio files in such a way that we can make inference? Do we segment the test audio files into windows and average the predictions on each window? Not to mention there is a possibility that there is no bird in each audio file.</p>",
      "rawMarkdown": "I am new to audio competitions. Sorry if this might seem like a silly / easy question.\n\nWe are given the labels for each 5-second window of each audio file. For inference, how do we process the 10-second audio files in such a way that we can make inference? Do we segment the test audio files into windows and average the predictions on each window? Not to mention there is a possibility that there is no bird in each audio file.",
      "votes": null
    },
    {
      "id": "1277421",
      "postDate": "04/18/2021 17:50:32",
      "content": "<p>There are examples of how to do it in the public notebooks.  Just open the code tab.</p>",
      "rawMarkdown": "There are examples of how to do it in the public notebooks.  Just open the code tab.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1277421,
      "author_name": "cpmpml",
      "author_url": "",
      "post_date": "04/18/2021 17:50:32",
      "content": "<p>There are examples of how to do it in the public notebooks.  Just open the code tab.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1277175": "I am new to audio competitions. Sorry if this might seem like a silly / easy question.\n\nWe are given the labels for each 5-second window of each audio file. For inference, how do we process the 10-second audio files in such a way that we can make inference? Do we segment the test audio files into windows and average the predictions on each window? Not to mention there is a possibility that there is no bird in each audio file.",
    "1277421": "There are examples of how to do it in the public notebooks.  Just open the code tab."
  },
  "source": "meta"
}