{
  "id": 169916,
  "title": "A Welcome Thread",
  "url": "/competitions/silero-audio-classifier/discussion/169916",
  "author_name": "",
  "post_date": "2020-07-25T16:58:58.223020Z",
  "votes": 3,
  "comment_count": 4,
  "views": 0,
  "content": "<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F851584%2F6da18c3fcfcf39ad7e079cbebdeeff89%2Fheader.png?generation=1595528284928601&amp;alt=media\" alt=\"\"></p>\n\n<p>Hi,</p>\n\n<p>This is a toy competition we uploaded for reasons =)</p>\n\n<p><strong>Anyway, you can find this interesting if:</strong></p>\n\n<ul>\n<li>You need to build a (voice activity detector) VAD;</li>\n<li>You are new to ML in general and you need a toy task to practice;</li>\n<li>You are new to sound processing and you need some model task to hone your skills;</li>\n<li>You are looking for some toy task to benchmark some alrgorithms;</li>\n</ul>\n\n<p><strong>Some reasoning behind the task:</strong></p>\n\n<ul>\n<li>VAD is a real production task, albeit in this case the task is over-simplified, i.e. you are not required to classify each audio frame, only entire files;</li>\n<li>We wanted to remove as much engineering / real world messiness from the data as much as possible;</li>\n<li>The data is very diverse, but the task is quite simple. There is a twist in the validation;</li>\n<li>It is quite possible that you could achieve high results (upwards of 95-97% accuracy), but it is worth exploring some simple ML methods as well just for fun;</li>\n</ul>\n\n<p>Make sure to take a look at the provided notebook(s)!\nHave fun! Enjoy, apply yourself!</p>",
  "messages": [
    {
      "id": "945225",
      "postDate": "07/25/2020 16:58:58",
      "content": "<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F851584%2F6da18c3fcfcf39ad7e079cbebdeeff89%2Fheader.png?generation=1595528284928601&amp;alt=media\" alt=\"\"></p>\n\n<p>Hi,</p>\n\n<p>This is a toy competition we uploaded for reasons =)</p>\n\n<p><strong>Anyway, you can find this interesting if:</strong></p>\n\n<ul>\n<li>You need to build a (voice activity detector) VAD;</li>\n<li>You are new to ML in general and you need a toy task to practice;</li>\n<li>You are new to sound processing and you need some model task to hone your skills;</li>\n<li>You are looking for some toy task to benchmark some alrgorithms;</li>\n</ul>\n\n<p><strong>Some reasoning behind the task:</strong></p>\n\n<ul>\n<li>VAD is a real production task, albeit in this case the task is over-simplified, i.e. you are not required to classify each audio frame, only entire files;</li>\n<li>We wanted to remove as much engineering / real world messiness from the data as much as possible;</li>\n<li>The data is very diverse, but the task is quite simple. There is a twist in the validation;</li>\n<li>It is quite possible that you could achieve high results (upwards of 95-97% accuracy), but it is worth exploring some simple ML methods as well just for fun;</li>\n</ul>\n\n<p>Make sure to take a look at the provided notebook(s)!\nHave fun! Enjoy, apply yourself!</p>",
      "rawMarkdown": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F851584%2F6da18c3fcfcf39ad7e079cbebdeeff89%2Fheader.png?generation=1595528284928601&amp;alt=media)\n\nHi,\n\nThis is a toy competition we uploaded for reasons =)\n\n**Anyway, you can find this interesting if:**\n\n- You need to build a (voice activity detector) VAD;\n- You are new to ML in general and you need a toy task to practice;\n- You are new to sound processing and you need some model task to hone your skills;\n- You are looking for some toy task to benchmark some alrgorithms;\n\n**Some reasoning behind the task:**\n\n- VAD is a real production task, albeit in this case the task is over-simplified, i.e. you are not required to classify each audio frame, only entire files;\n- We wanted to remove as much engineering / real world messiness from the data as much as possible;\n- The data is very diverse, but the task is quite simple. There is a twist in the validation;\n- It is quite possible that you could achieve high results (upwards of 95-97% accuracy), but it is worth exploring some simple ML methods as well just for fun;\n\nMake sure to take a look at the provided notebook(s)!\nHave fun! Enjoy, apply yourself!",
      "votes": null
    },
    {
      "id": "949671",
      "postDate": "07/28/2020 19:41:11",
      "content": "<p>am stack. any help guys?</p>",
      "rawMarkdown": "am stack. any help guys?",
      "votes": null
    },
    {
      "id": "949887",
      "postDate": "07/29/2020 03:30:49",
      "content": "<p>What exactly is your problem?</p>",
      "rawMarkdown": "What exactly is your problem?",
      "votes": null
    },
    {
      "id": "951340",
      "postDate": "07/30/2020 04:40:17",
      "content": "<p>got fixed got fixed am sorry, thanks</p>",
      "rawMarkdown": "got fixed got fixed am sorry, thanks",
      "votes": null
    },
    {
      "id": "952426",
      "postDate": "07/30/2020 22:48:13",
      "content": "<p>hey hello, am getting a 'in user code' error, weird thou'!</p>",
      "rawMarkdown": "hey hello, am getting a 'in user code' error, weird thou'!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 949671,
      "author_name": "gicehajunior",
      "author_url": "",
      "post_date": "07/28/2020 19:41:11",
      "content": "<p>am stack. any help guys?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 949887,
      "author_name": "snakers41",
      "author_url": "",
      "post_date": "07/29/2020 03:30:49",
      "content": "<p>What exactly is your problem?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 951340,
      "author_name": "gicehajunior",
      "author_url": "",
      "post_date": "07/30/2020 04:40:17",
      "content": "<p>got fixed got fixed am sorry, thanks</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 952426,
      "author_name": "gicehajunior",
      "author_url": "",
      "post_date": "07/30/2020 22:48:13",
      "content": "<p>hey hello, am getting a 'in user code' error, weird thou'!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "945225": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F851584%2F6da18c3fcfcf39ad7e079cbebdeeff89%2Fheader.png?generation=1595528284928601&amp;alt=media)\n\nHi,\n\nThis is a toy competition we uploaded for reasons =)\n\n**Anyway, you can find this interesting if:**\n\n- You need to build a (voice activity detector) VAD;\n- You are new to ML in general and you need a toy task to practice;\n- You are new to sound processing and you need some model task to hone your skills;\n- You are looking for some toy task to benchmark some alrgorithms;\n\n**Some reasoning behind the task:**\n\n- VAD is a real production task, albeit in this case the task is over-simplified, i.e. you are not required to classify each audio frame, only entire files;\n- We wanted to remove as much engineering / real world messiness from the data as much as possible;\n- The data is very diverse, but the task is quite simple. There is a twist in the validation;\n- It is quite possible that you could achieve high results (upwards of 95-97% accuracy), but it is worth exploring some simple ML methods as well just for fun;\n\nMake sure to take a look at the provided notebook(s)!\nHave fun! Enjoy, apply yourself!",
    "949671": "am stack. any help guys?",
    "949887": "What exactly is your problem?",
    "951340": "got fixed got fixed am sorry, thanks",
    "952426": "hey hello, am getting a 'in user code' error, weird thou'!"
  },
  "source": "meta"
}