{
  "id": 454151,
  "title": "Question about external data on Video setting",
  "url": "/competitions/audio-driven-talking-head-generation/discussion/454151",
  "author_name": "",
  "post_date": "2023-11-09T06:33:28.647872Z",
  "votes": 1,
  "comment_count": 1,
  "views": 0,
  "content": "<p>As your metioned on the Video setting, we are confused about the defination about external data. Could you please help us to answer these question?</p>\n<p>May I use the pretrained models (like pretrained LLM) as audio encoder?<br>\nMay I use the pretrained models trained from HDTF and finetune on this task?</p>",
  "messages": [
    {
      "id": "2518209",
      "postDate": "11/09/2023 06:33:28",
      "content": "<p>As your metioned on the Video setting, we are confused about the defination about external data. Could you please help us to answer these question?</p>\n<p>May I use the pretrained models (like pretrained LLM) as audio encoder?<br>\nMay I use the pretrained models trained from HDTF and finetune on this task?</p>",
      "rawMarkdown": "As your metioned on the Video setting, we are confused about the defination about external data. Could you please help us to answer these question?\n\nMay I use the pretrained models (like pretrained LLM) as audio encoder?\nMay I use the pretrained models trained from HDTF and finetune on this task?",
      "votes": null
    },
    {
      "id": "2518319",
      "postDate": "11/09/2023 07:43:12",
      "content": "<p>Thank you for your interest in the video setting and thanks for the comment. We have discussed about the usage of external data and the reply would be as follows:</p>\n<ol>\n<li>Participants are allowed to use pretrained audio models to process audios. </li>\n<li>Pariticpants are not allowed to use pretrained talking head(visual) model and finetune under the video setting.</li>\n</ol>\n<p>Since this track mainly focus on the generation quality of talking head instead of the audio itself, it is allowed to use pretrained models as audio encoder. However, we encourage participants to develop an end-to-end pipeline for this task, so pretrained models trained from HDTF are not allowed.</p>",
      "rawMarkdown": "Thank you for your interest in the video setting and thanks for the comment. We have discussed about the usage of external data and the reply would be as follows:\n1. Participants are allowed to use pretrained audio models to process audios. \n2. Pariticpants are not allowed to use pretrained talking head(visual) model and finetune under the video setting.\n\nSince this track mainly focus on the generation quality of talking head instead of the audio itself, it is allowed to use pretrained models as audio encoder. However, we encourage participants to develop an end-to-end pipeline for this task, so pretrained models trained from HDTF are not allowed.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2518319,
      "author_name": "jingnangao",
      "author_url": "",
      "post_date": "11/09/2023 07:43:12",
      "content": "<p>Thank you for your interest in the video setting and thanks for the comment. We have discussed about the usage of external data and the reply would be as follows:</p>\n<ol>\n<li>Participants are allowed to use pretrained audio models to process audios. </li>\n<li>Pariticpants are not allowed to use pretrained talking head(visual) model and finetune under the video setting.</li>\n</ol>\n<p>Since this track mainly focus on the generation quality of talking head instead of the audio itself, it is allowed to use pretrained models as audio encoder. However, we encourage participants to develop an end-to-end pipeline for this task, so pretrained models trained from HDTF are not allowed.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2518209": "As your metioned on the Video setting, we are confused about the defination about external data. Could you please help us to answer these question?\n\nMay I use the pretrained models (like pretrained LLM) as audio encoder?\nMay I use the pretrained models trained from HDTF and finetune on this task?",
    "2518319": "Thank you for your interest in the video setting and thanks for the comment. We have discussed about the usage of external data and the reply would be as follows:\n1. Participants are allowed to use pretrained audio models to process audios. \n2. Pariticpants are not allowed to use pretrained talking head(visual) model and finetune under the video setting.\n\nSince this track mainly focus on the generation quality of talking head instead of the audio itself, it is allowed to use pretrained models as audio encoder. However, we encourage participants to develop an end-to-end pipeline for this task, so pretrained models trained from HDTF are not allowed."
  },
  "source": "meta"
}