{
  "id": 129015,
  "title": "We need to read audio from the files.",
  "url": "/competitions/deepfake-detection-challenge/discussion/129015",
  "author_name": "",
  "post_date": "2020-02-04T21:19:22.653266900Z",
  "votes": null,
  "comment_count": 6,
  "views": 0,
  "content": "<p>We all need something to get audio. Something like MoviePY, FFMPEG or anything really.\nHello kaggle team. \nI feel this isn't a hacking competition. But a code competition.\nWe've spent time on it and asked many times about something in docker.\nWe NEED something for audio.</p>",
  "messages": [
    {
      "id": "737050",
      "postDate": "02/04/2020 21:19:22",
      "content": "<p>We all need something to get audio. Something like MoviePY, FFMPEG or anything really.\nHello kaggle team. \nI feel this isn't a hacking competition. But a code competition.\nWe've spent time on it and asked many times about something in docker.\nWe NEED something for audio.</p>",
      "rawMarkdown": "We all need something to get audio. Something like MoviePY, FFMPEG or anything really.\nHello kaggle team. \nI feel this isn't a hacking competition. But a code competition.\nWe've spent time on it and asked many times about something in docker.\nWe NEED something for audio.",
      "votes": null
    },
    {
      "id": "737994",
      "postDate": "02/06/2020 02:38:26",
      "content": "<p>I've got audio working with ffmpeg. Someone else uploaded a ffmpeg static build that you can use relatively painlessly. It is shown in a few public kernels I believe. The way I am doing it is unzip the ffmpeg folder then call the ffmpeg command line commands using subprocess library in python. That outputs a file that I can open up in librosa</p>",
      "rawMarkdown": "I've got audio working with ffmpeg. Someone else uploaded a ffmpeg static build that you can use relatively painlessly. It is shown in a few public kernels I believe. The way I am doing it is unzip the ffmpeg folder then call the ffmpeg command line commands using subprocess library in python. That outputs a file that I can open up in librosa",
      "votes": null
    },
    {
      "id": "738435",
      "postDate": "02/06/2020 14:16:54",
      "content": "<p>I feel like uploading an executable on a distant machine and running it in a shell is bit odd. Feels like the limits are blurred around the rules.\nA docker with the right simple tools could have spared us these funny pieces of code we're all writing now. But we'll be fine at then end !</p>",
      "rawMarkdown": "I feel like uploading an executable on a distant machine and running it in a shell is bit odd. Feels like the limits are blurred around the rules.\nA docker with the right simple tools could have spared us these funny pieces of code we're all writing now. But we'll be fine at then end !",
      "votes": null
    },
    {
      "id": "746715",
      "postDate": "02/15/2020 12:58:34",
      "content": "<p>Hi,ryches. It seems that we use the same way to process the audio files. But I met some problem while using ffmpeg. Here is my code:<code>command = f\"ffmpeg-git-20200114-amd64-static/ffmpeg -i {file} -ab 192000 -ac 2 -ar 44100 -vn {file[-14:-4]}.{output_format}\"</code> and <code>subprocess.call(command, shell=True)</code>. Can you show me the bug in my code? Maybe I wrote wrong paths? Thanks!</p>",
      "rawMarkdown": "Hi,ryches. It seems that we use the same way to process the audio files. But I met some problem while using ffmpeg. Here is my code:`command = f\"ffmpeg-git-20200114-amd64-static/ffmpeg -i {file} -ab 192000 -ac 2 -ar 44100 -vn {file[-14:-4]}.{output_format}\"` and `subprocess.call(command, shell=True)`. Can you show me the bug in my code? Maybe I wrote wrong paths? Thanks!",
      "votes": null
    },
    {
      "id": "746725",
      "postDate": "02/15/2020 13:10:34",
      "content": "<p>oh, I remember I did a test submit. I found my code couldnt find the wav files produced by ffmpeg. I dont know whether it is because I didnt build the wav file or I just didnt write the right path to wav files. </p>",
      "rawMarkdown": "oh, I remember I did a test submit. I found my code couldnt find the wav files produced by ffmpeg. I dont know whether it is because I didnt build the wav file or I just didnt write the right path to wav files.",
      "votes": null
    },
    {
      "id": "760285",
      "postDate": "03/01/2020 04:24:59",
      "content": "<p>hi,<a href=\"/huangliuyu\">@huangliuyu</a>, we have the same problem with you. Have you already solve the problem? We can run on the trainsample, but got \"connot find the CSV file\" error after submit it.</p>",
      "rawMarkdown": "hi,@huangliuyu, we have the same problem with you. Have you already solve the problem? We can run on the trainsample, but got \"connot find the CSV file\" error after submit it.",
      "votes": null
    },
    {
      "id": "761438",
      "postDate": "03/02/2020 14:19:12",
      "content": "<p>Hi,<a href=\"/huangchaoqin\">@huangchaoqin</a>，I'm a teammate of huangliuyu. We finally solved the problem. But our problem is  that we cant find the .wav file. If you cant find the output csv file, how did you get the 121st for now... But actually, the files created by us  were found in the path of \"../working/\". If you are interested in our model,maybe we can team up.</p>",
      "rawMarkdown": "Hi,@huangchaoqin，I'm a teammate of huangliuyu. We finally solved the problem. But our problem is  that we cant find the .wav file. If you cant find the output csv file, how did you get the 121st for now... But actually, the files created by us  were found in the path of \"../working/\". If you are interested in our model,maybe we can team up.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 737994,
      "author_name": "ryches",
      "author_url": "",
      "post_date": "02/06/2020 02:38:26",
      "content": "<p>I've got audio working with ffmpeg. Someone else uploaded a ffmpeg static build that you can use relatively painlessly. It is shown in a few public kernels I believe. The way I am doing it is unzip the ffmpeg folder then call the ffmpeg command line commands using subprocess library in python. That outputs a file that I can open up in librosa</p>",
      "votes": null,
      "replies": [
        {
          "id": 746715,
          "author_name": "huangliuyu",
          "author_url": "",
          "post_date": "02/15/2020 12:58:34",
          "content": "<p>Hi,ryches. It seems that we use the same way to process the audio files. But I met some problem while using ffmpeg. Here is my code:<code>command = f\"ffmpeg-git-20200114-amd64-static/ffmpeg -i {file} -ab 192000 -ac 2 -ar 44100 -vn {file[-14:-4]}.{output_format}\"</code> and <code>subprocess.call(command, shell=True)</code>. Can you show me the bug in my code? Maybe I wrote wrong paths? Thanks!</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 746725,
          "author_name": "huangliuyu",
          "author_url": "",
          "post_date": "02/15/2020 13:10:34",
          "content": "<p>oh, I remember I did a test submit. I found my code couldnt find the wav files produced by ffmpeg. I dont know whether it is because I didnt build the wav file or I just didnt write the right path to wav files. </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 760285,
          "author_name": "huangchaoqin",
          "author_url": "",
          "post_date": "03/01/2020 04:24:59",
          "content": "<p>hi,<a href=\"/huangliuyu\">@huangliuyu</a>, we have the same problem with you. Have you already solve the problem? We can run on the trainsample, but got \"connot find the CSV file\" error after submit it.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 761438,
          "author_name": "emmettj",
          "author_url": "",
          "post_date": "03/02/2020 14:19:12",
          "content": "<p>Hi,<a href=\"/huangchaoqin\">@huangchaoqin</a>，I'm a teammate of huangliuyu. We finally solved the problem. But our problem is  that we cant find the .wav file. If you cant find the output csv file, how did you get the 121st for now... But actually, the files created by us  were found in the path of \"../working/\". If you are interested in our model,maybe we can team up.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 738435,
      "author_name": "simoninparis",
      "author_url": "",
      "post_date": "02/06/2020 14:16:54",
      "content": "<p>I feel like uploading an executable on a distant machine and running it in a shell is bit odd. Feels like the limits are blurred around the rules.\nA docker with the right simple tools could have spared us these funny pieces of code we're all writing now. But we'll be fine at then end !</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "737050": "We all need something to get audio. Something like MoviePY, FFMPEG or anything really.\nHello kaggle team. \nI feel this isn't a hacking competition. But a code competition.\nWe've spent time on it and asked many times about something in docker.\nWe NEED something for audio.",
    "737994": "I've got audio working with ffmpeg. Someone else uploaded a ffmpeg static build that you can use relatively painlessly. It is shown in a few public kernels I believe. The way I am doing it is unzip the ffmpeg folder then call the ffmpeg command line commands using subprocess library in python. That outputs a file that I can open up in librosa",
    "738435": "I feel like uploading an executable on a distant machine and running it in a shell is bit odd. Feels like the limits are blurred around the rules.\nA docker with the right simple tools could have spared us these funny pieces of code we're all writing now. But we'll be fine at then end !",
    "746715": "Hi,ryches. It seems that we use the same way to process the audio files. But I met some problem while using ffmpeg. Here is my code:`command = f\"ffmpeg-git-20200114-amd64-static/ffmpeg -i {file} -ab 192000 -ac 2 -ar 44100 -vn {file[-14:-4]}.{output_format}\"` and `subprocess.call(command, shell=True)`. Can you show me the bug in my code? Maybe I wrote wrong paths? Thanks!",
    "746725": "oh, I remember I did a test submit. I found my code couldnt find the wav files produced by ffmpeg. I dont know whether it is because I didnt build the wav file or I just didnt write the right path to wav files.",
    "760285": "hi,@huangliuyu, we have the same problem with you. Have you already solve the problem? We can run on the trainsample, but got \"connot find the CSV file\" error after submit it.",
    "761438": "Hi,@huangchaoqin，I'm a teammate of huangliuyu. We finally solved the problem. But our problem is  that we cant find the .wav file. If you cant find the output csv file, how did you get the 121st for now... But actually, the files created by us  were found in the path of \"../working/\". If you are interested in our model,maybe we can team up."
  },
  "source": "meta"
}