{
  "id": 569105,
  "title": "Same url for multiple species: audio including multiple birds species?",
  "url": "/competitions/birdclef-2025/discussion/569105",
  "author_name": "",
  "post_date": "2025-03-20T01:29:26.560234400Z",
  "votes": 8,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi guys, I just wanna to share a small discovery here. Some of the samples with different label are having the same url, is that mean the audio includes birds calls from multiple birds species at once? Although these three file (and there are several others) have different sha256sum, I listened and can confirm they are the same. Is removing them from the training set a good idea? (Assuming using random 5sec sample)</p>\n<pre><code>      greibi1             [\n      pavpig2             [\n      pirfly1             [\n</code></pre>",
  "messages": [
    {
      "id": "3154465",
      "postDate": "03/20/2025 01:29:26",
      "content": "<p>Hi guys, I just wanna to share a small discovery here. Some of the samples with different label are having the same url, is that mean the audio includes birds calls from multiple birds species at once? Although these three file (and there are several others) have different sha256sum, I listened and can confirm they are the same. Is removing them from the training set a good idea? (Assuming using random 5sec sample)</p>\n<pre><code>      greibi1             [\n      pavpig2             [\n      pirfly1             [\n</code></pre>",
      "rawMarkdown": "Hi guys, I just wanna to share a small discovery here. Some of the samples with different label are having the same url, is that mean the audio includes birds calls from multiple birds species at once? Although these three file (and there are several others) have different sha256sum, I listened and can confirm they are the same. Is removing them from the training set a good idea? (Assuming using random 5sec sample)\n```\n10396       greibi1             [''] [''] greibi1/iNat1020465.ogg       iNat      0 https://static.inaturalist.org/sounds/1020465.wav   4.8329  -52.2985 Mesembrinibis cayennensis\n14430       pavpig2             [''] [''] pavpig2/iNat1020465.ogg       iNat      0 https://static.inaturalist.org/sounds/1020465.wav   4.8329  -52.2985   Patagioenas cayennensis\n14784       pirfly1             [''] [''] pirfly1/iNat1020465.ogg       iNat      0 https://static.inaturalist.org/sounds/1020465.wav   4.8329  -52.2985       Legatus leucophaius\n```",
      "votes": null
    },
    {
      "id": "3154679",
      "postDate": "03/20/2025 09:16:13",
      "content": "<p>It is totally possible that these do actually contain all three species and were uploaded to iNat with three observations/species IDs.</p>",
      "rawMarkdown": "It is totally possible that these do actually contain all three species and were uploaded to iNat with three observations/species IDs.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3154679,
      "author_name": "stefankahl",
      "author_url": "",
      "post_date": "03/20/2025 09:16:13",
      "content": "<p>It is totally possible that these do actually contain all three species and were uploaded to iNat with three observations/species IDs.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3154465": "Hi guys, I just wanna to share a small discovery here. Some of the samples with different label are having the same url, is that mean the audio includes birds calls from multiple birds species at once? Although these three file (and there are several others) have different sha256sum, I listened and can confirm they are the same. Is removing them from the training set a good idea? (Assuming using random 5sec sample)\n```\n10396       greibi1             [''] [''] greibi1/iNat1020465.ogg       iNat      0 https://static.inaturalist.org/sounds/1020465.wav   4.8329  -52.2985 Mesembrinibis cayennensis\n14430       pavpig2             [''] [''] pavpig2/iNat1020465.ogg       iNat      0 https://static.inaturalist.org/sounds/1020465.wav   4.8329  -52.2985   Patagioenas cayennensis\n14784       pirfly1             [''] [''] pirfly1/iNat1020465.ogg       iNat      0 https://static.inaturalist.org/sounds/1020465.wav   4.8329  -52.2985       Legatus leucophaius\n```",
    "3154679": "It is totally possible that these do actually contain all three species and were uploaded to iNat with three observations/species IDs."
  },
  "source": "meta"
}