{
  "id": 394047,
  "title": "My Solutions - Texts Classification and sounds Recognition",
  "url": "/competitions/ml-olympiad-dialectrecognition/discussion/394047",
  "author_name": "",
  "post_date": "2023-03-11T21:25:50.829528Z",
  "votes": 2,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi everyone,</p>\n<p>I used several technique to recognize dialect (Text classification and sound Recognition) during this competition </p>\n<p><strong>Text Classification model:</strong><br>\nI used Keras library to prepare data and build the model. This notebook include implementation details for the keras model that was trained for dialects classification. <a href=\"https://www.kaggle.com/code/zarahshibli/keras-model-to-classify-speaker-dialect?scriptVersionId=121800651\" target=\"_blank\">Keras model to classify speaker dialect</a>.</p>\n<p><strong>Sound Recognition model:</strong> <br>\nFirstly, I created sub dataset, by splitting audio to segments, each segment start from SegmentStart and end by SegmentEnd. I have named output file from previous step by segmentID.</p>\n<p>Secondly, I extracted features from audio using librosa. Then build a simple Keras model with 4 dense layers. To see more details about the implementation, you can check this notebook <a href=\"https://www.kaggle.com/code/zarahshibli/speaker-sound-dialect-classification-using-keras?scriptVersionId=121804434\" target=\"_blank\">Speaker sound dialect classification using Keras</a>.</p>\n<p><strong>Challenges:</strong></p>\n<ol>\n<li><p>While working in this competition I had many challenges. I can summarize it in the following points:</p></li>\n<li><p>Extract features from audio take long time.<br>\nI encountered an error running the code in Kaggle when I tried to increase the number of audio files. Such as \"File contains data in an unknown format\".</p></li>\n</ol>\n<p>I wish to thank SDAIA and SBA for giving us the opportunity to work on the data. </p>\n<p>Finally, I would like to thank the organizers of this competition. </p>",
  "messages": [
    {
      "id": "2177864",
      "postDate": "03/11/2023 21:25:50",
      "content": "<p>Hi everyone,</p>\n<p>I used several technique to recognize dialect (Text classification and sound Recognition) during this competition </p>\n<p><strong>Text Classification model:</strong><br>\nI used Keras library to prepare data and build the model. This notebook include implementation details for the keras model that was trained for dialects classification. <a href=\"https://www.kaggle.com/code/zarahshibli/keras-model-to-classify-speaker-dialect?scriptVersionId=121800651\" target=\"_blank\">Keras model to classify speaker dialect</a>.</p>\n<p><strong>Sound Recognition model:</strong> <br>\nFirstly, I created sub dataset, by splitting audio to segments, each segment start from SegmentStart and end by SegmentEnd. I have named output file from previous step by segmentID.</p>\n<p>Secondly, I extracted features from audio using librosa. Then build a simple Keras model with 4 dense layers. To see more details about the implementation, you can check this notebook <a href=\"https://www.kaggle.com/code/zarahshibli/speaker-sound-dialect-classification-using-keras?scriptVersionId=121804434\" target=\"_blank\">Speaker sound dialect classification using Keras</a>.</p>\n<p><strong>Challenges:</strong></p>\n<ol>\n<li><p>While working in this competition I had many challenges. I can summarize it in the following points:</p></li>\n<li><p>Extract features from audio take long time.<br>\nI encountered an error running the code in Kaggle when I tried to increase the number of audio files. Such as \"File contains data in an unknown format\".</p></li>\n</ol>\n<p>I wish to thank SDAIA and SBA for giving us the opportunity to work on the data. </p>\n<p>Finally, I would like to thank the organizers of this competition. </p>",
      "rawMarkdown": "Hi everyone,\n\n\nI used several technique to recognize dialect (Text classification and sound Recognition) during this competition \n\n**Text Classification model:**\nI used Keras library to prepare data and build the model. This notebook include implementation details for the keras model that was trained for dialects classification. [Keras model to classify speaker dialect](https://www.kaggle.com/code/zarahshibli/keras-model-to-classify-speaker-dialect?scriptVersionId=121800651).\n\n**Sound Recognition model:** \nFirstly, I created sub dataset, by splitting audio to segments, each segment start from SegmentStart and end by SegmentEnd. I have named output file from previous step by segmentID.\n\nSecondly, I extracted features from audio using librosa. Then build a simple Keras model with 4 dense layers. To see more details about the implementation, you can check this notebook [Speaker sound dialect classification using Keras](https://www.kaggle.com/code/zarahshibli/speaker-sound-dialect-classification-using-keras?scriptVersionId=121804434).\n\n\n**Challenges:**\n1. While working in this competition I had many challenges. I can summarize it in the following points:\n\n2. Extract features from audio take long time.\nI encountered an error running the code in Kaggle when I tried to increase the number of audio files. Such as \"File contains data in an unknown format\".\n\n\nI wish to thank SDAIA and SBA for giving us the opportunity to work on the data. \n\nFinally, I would like to thank the organizers of this competition.",
      "votes": null
    },
    {
      "id": "2178054",
      "postDate": "03/12/2023 03:53:59",
      "content": "<p>Thanks for sharing! <br>\nBest Wishes and happy learning </p>",
      "rawMarkdown": "Thanks for sharing! \nBest Wishes and happy learning",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2178054,
      "author_name": "asalhi",
      "author_url": "",
      "post_date": "03/12/2023 03:53:59",
      "content": "<p>Thanks for sharing! <br>\nBest Wishes and happy learning </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2177864": "Hi everyone,\n\n\nI used several technique to recognize dialect (Text classification and sound Recognition) during this competition \n\n**Text Classification model:**\nI used Keras library to prepare data and build the model. This notebook include implementation details for the keras model that was trained for dialects classification. [Keras model to classify speaker dialect](https://www.kaggle.com/code/zarahshibli/keras-model-to-classify-speaker-dialect?scriptVersionId=121800651).\n\n**Sound Recognition model:** \nFirstly, I created sub dataset, by splitting audio to segments, each segment start from SegmentStart and end by SegmentEnd. I have named output file from previous step by segmentID.\n\nSecondly, I extracted features from audio using librosa. Then build a simple Keras model with 4 dense layers. To see more details about the implementation, you can check this notebook [Speaker sound dialect classification using Keras](https://www.kaggle.com/code/zarahshibli/speaker-sound-dialect-classification-using-keras?scriptVersionId=121804434).\n\n\n**Challenges:**\n1. While working in this competition I had many challenges. I can summarize it in the following points:\n\n2. Extract features from audio take long time.\nI encountered an error running the code in Kaggle when I tried to increase the number of audio files. Such as \"File contains data in an unknown format\".\n\n\nI wish to thank SDAIA and SBA for giving us the opportunity to work on the data. \n\nFinally, I would like to thank the organizers of this competition.",
    "2178054": "Thanks for sharing! \nBest Wishes and happy learning"
  },
  "source": "meta"
}