{
  "id": 201929,
  "title": "[Datasets] Megathread for Datasets for faster training/inference",
  "url": "/competitions/rfcx-species-audio-detection/discussion/201929",
  "author_name": "Ultron",
  "post_date": "2020-12-07T10:43:08.547000",
  "votes": 4,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I believe we need to save the possible formats, splits of the datasets for this competition as it takes too much time to load the audio files as well as process them in the run time. It will save a hell amount of a time. Here I am posting my recent work on the dataset and features engineering.</p>\n<h1>1. <a href=\"https://www.kaggle.com/mrutyunjaybiswal/mfccrfcx\" target=\"_blank\">MFCC and beyond</a></h1>\n<h3>Content</h3>\n<p>As of now, it contains the stacking of the following features:</p>\n<ul>\n<li>harmonics and percussive</li>\n<li>Statis and Dynamic MFCCs</li>\n<li>Chroma features</li>\n</ul>\n<p>The shape of each file: (74, 164, 1)<br>\nOf course, the last dimension is added to make it 3 dim only.</p>\n<p>Notebook: <a href=\"https://www.kaggle.com/mrutyunjaybiswal/rainforest-features-dataset-mfcc-and-beyond\" target=\"_blank\">https://www.kaggle.com/mrutyunjaybiswal/rainforest-features-dataset-mfcc-and-beyond</a></p>\n<p>NOTE: Updated versions, as well as different features, will be added as the competition and my work progresses. Hope it helps, do provide feedback as well as corrections wherever required. Thanks ;)</p>\n<p>I hope others will add their respective datasets with extracted features to get started as well. Thanks in advance.</p>",
  "messages": [
    {
      "id": 1104897,
      "postDate": "2020-12-07T10:43:08.547Z",
      "content": "<p>I believe we need to save the possible formats, splits of the datasets for this competition as it takes too much time to load the audio files as well as process them in the run time. It will save a hell amount of a time. Here I am posting my recent work on the dataset and features engineering.</p>\n<h1>1. <a href=\"https://www.kaggle.com/mrutyunjaybiswal/mfccrfcx\" target=\"_blank\">MFCC and beyond</a></h1>\n<h3>Content</h3>\n<p>As of now, it contains the stacking of the following features:</p>\n<ul>\n<li>harmonics and percussive</li>\n<li>Statis and Dynamic MFCCs</li>\n<li>Chroma features</li>\n</ul>\n<p>The shape of each file: (74, 164, 1)<br>\nOf course, the last dimension is added to make it 3 dim only.</p>\n<p>Notebook: <a href=\"https://www.kaggle.com/mrutyunjaybiswal/rainforest-features-dataset-mfcc-and-beyond\" target=\"_blank\">https://www.kaggle.com/mrutyunjaybiswal/rainforest-features-dataset-mfcc-and-beyond</a></p>\n<p>NOTE: Updated versions, as well as different features, will be added as the competition and my work progresses. Hope it helps, do provide feedback as well as corrections wherever required. Thanks ;)</p>\n<p>I hope others will add their respective datasets with extracted features to get started as well. Thanks in advance.</p>",
      "rawMarkdown": "I believe we need to save the possible formats, splits of the datasets for this competition as it takes too much time to load the audio files as well as process them in the run time. It will save a hell amount of a time. Here I am posting my recent work on the dataset and features engineering.\n\n# 1. [MFCC and beyond](https://www.kaggle.com/mrutyunjaybiswal/mfccrfcx) \n\n### Content\n\nAs of now, it contains the stacking of the following features:\n\n- harmonics and percussive\n- Statis and Dynamic MFCCs\n- Chroma features\n\nThe shape of each file: (74, 164, 1)\nOf course, the last dimension is added to make it 3 dim only.\n\nNotebook: https://www.kaggle.com/mrutyunjaybiswal/rainforest-features-dataset-mfcc-and-beyond\n\nNOTE: Updated versions, as well as different features, will be added as the competition and my work progresses. Hope it helps, do provide feedback as well as corrections wherever required. Thanks ;)\n\nI hope others will add their respective datasets with extracted features to get started as well. Thanks in advance.",
      "votes": 4
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1104897": "I believe we need to save the possible formats, splits of the datasets for this competition as it takes too much time to load the audio files as well as process them in the run time. It will save a hell amount of a time. Here I am posting my recent work on the dataset and features engineering.\n\n# 1. [MFCC and beyond](https://www.kaggle.com/mrutyunjaybiswal/mfccrfcx) \n\n### Content\n\nAs of now, it contains the stacking of the following features:\n\n- harmonics and percussive\n- Statis and Dynamic MFCCs\n- Chroma features\n\nThe shape of each file: (74, 164, 1)\nOf course, the last dimension is added to make it 3 dim only.\n\nNotebook: https://www.kaggle.com/mrutyunjaybiswal/rainforest-features-dataset-mfcc-and-beyond\n\nNOTE: Updated versions, as well as different features, will be added as the competition and my work progresses. Hope it helps, do provide feedback as well as corrections wherever required. Thanks ;)\n\nI hope others will add their respective datasets with extracted features to get started as well. Thanks in advance."
  }
}