{
  "id": 177177,
  "title": "Learning Path for speech recognistion",
  "url": "/competitions/birdsong-recognition/discussion/177177",
  "author_name": "",
  "post_date": "2020-08-25T04:56:15.031868900Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>I am new to speech recognistion can any body guide me how to get started</p>",
  "messages": [
    {
      "id": "984415",
      "postDate": "08/25/2020 04:56:15",
      "content": "<p>I am new to speech recognistion can any body guide me how to get started</p>",
      "rawMarkdown": "I am new to speech recognistion can any body guide me how to get started",
      "votes": null
    },
    {
      "id": "984625",
      "postDate": "08/25/2020 07:33:43",
      "content": "<p>I will tell you briefly… (what most people are doing in notebooks here and elsewhere)</p>\n<p>You have some audio files for each bird - given data. </p>\n<p>First, you have to convert these into a spectrogram/melspectogram. This is an image of a given piece of sound.</p>\n<p>A bit detail of this process - This article can help<br>\n<a href=\"https://medium.com/@ageitgey/machine-learning-is-fun-part-6-how-to-do-speech-recognition-with-deep-learning-28293c162f7a\" target=\"_blank\">https://medium.com/@ageitgey/machine-learning-is-fun-part-6-how-to-do-speech-recognition-with-deep-learning-28293c162f7a</a></p>\n<p>Use librosa a python library to do that. (very easy to use) - This article can help<br>\n<a href=\"https://heartbeat.fritz.ai/working-with-audio-signals-in-python-6c2bd63b2daf\" target=\"_blank\">https://heartbeat.fritz.ai/working-with-audio-signals-in-python-6c2bd63b2daf</a></p>\n<p>So you get some spectograms/melspectograms for each bird - processed data</p>\n<p>Second, you use image processing (like Resnet etc..,) to train models on those images (each bird is a class)</p>\n<p>During prediction - same procedure…<br>\nconvert given audio files into spectograms/melspectograms<br>\npredict using your trained model</p>\n<p>Mostly, that's it.</p>",
      "rawMarkdown": "I will tell you briefly... (what most people are doing in notebooks here and elsewhere)\n\nYou have some audio files for each bird - given data. \n\nFirst, you have to convert these into a spectrogram/melspectogram. This is an image of a given piece of sound.\n\nA bit detail of this process - This article can help\nhttps://medium.com/@ageitgey/machine-learning-is-fun-part-6-how-to-do-speech-recognition-with-deep-learning-28293c162f7a\n\nUse librosa a python library to do that. (very easy to use) - This article can help\nhttps://heartbeat.fritz.ai/working-with-audio-signals-in-python-6c2bd63b2daf\n\nSo you get some spectograms/melspectograms for each bird - processed data\n\nSecond, you use image processing (like Resnet etc..,) to train models on those images (each bird is a class)\n\nDuring prediction - same procedure...\nconvert given audio files into spectograms/melspectograms\npredict using your trained model\n\nMostly, that's it.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 984625,
      "author_name": "vsvrp1995",
      "author_url": "",
      "post_date": "08/25/2020 07:33:43",
      "content": "<p>I will tell you briefly… (what most people are doing in notebooks here and elsewhere)</p>\n<p>You have some audio files for each bird - given data. </p>\n<p>First, you have to convert these into a spectrogram/melspectogram. This is an image of a given piece of sound.</p>\n<p>A bit detail of this process - This article can help<br>\n<a href=\"https://medium.com/@ageitgey/machine-learning-is-fun-part-6-how-to-do-speech-recognition-with-deep-learning-28293c162f7a\" target=\"_blank\">https://medium.com/@ageitgey/machine-learning-is-fun-part-6-how-to-do-speech-recognition-with-deep-learning-28293c162f7a</a></p>\n<p>Use librosa a python library to do that. (very easy to use) - This article can help<br>\n<a href=\"https://heartbeat.fritz.ai/working-with-audio-signals-in-python-6c2bd63b2daf\" target=\"_blank\">https://heartbeat.fritz.ai/working-with-audio-signals-in-python-6c2bd63b2daf</a></p>\n<p>So you get some spectograms/melspectograms for each bird - processed data</p>\n<p>Second, you use image processing (like Resnet etc..,) to train models on those images (each bird is a class)</p>\n<p>During prediction - same procedure…<br>\nconvert given audio files into spectograms/melspectograms<br>\npredict using your trained model</p>\n<p>Mostly, that's it.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "984415": "I am new to speech recognistion can any body guide me how to get started",
    "984625": "I will tell you briefly... (what most people are doing in notebooks here and elsewhere)\n\nYou have some audio files for each bird - given data. \n\nFirst, you have to convert these into a spectrogram/melspectogram. This is an image of a given piece of sound.\n\nA bit detail of this process - This article can help\nhttps://medium.com/@ageitgey/machine-learning-is-fun-part-6-how-to-do-speech-recognition-with-deep-learning-28293c162f7a\n\nUse librosa a python library to do that. (very easy to use) - This article can help\nhttps://heartbeat.fritz.ai/working-with-audio-signals-in-python-6c2bd63b2daf\n\nSo you get some spectograms/melspectograms for each bird - processed data\n\nSecond, you use image processing (like Resnet etc..,) to train models on those images (each bird is a class)\n\nDuring prediction - same procedure...\nconvert given audio files into spectograms/melspectograms\npredict using your trained model\n\nMostly, that's it."
  },
  "source": "meta"
}