You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

gol-dataset-2k

データセット構造

データインスタンス

各インスタンスには以下が含まれます:

{
  "audio": {
    "path": "folder_id/speaker_id/audio_file.wav",
    "sampling_rate": 48000
  },
  "text": "こんにちは、今日はいい天気ですね。",
  "speaker_id": "SPEAKER_HASH_ID",
  "duration": 2.5,
  "emotion": {
    "喜び": 0.85,
    "怒り": 0.02,
    "悲しみ": 0.01,
    "恐怖": 0.01,
    "驚き": 0.05,
    "嫌悪": 0.01,
    "楽しみ": 0.04,
    "ニュートラル": 0.01
  }
}

データフィールド

  • audio: WAV形式の音声ファイル
  • text: 発話内容の書き起こし
  • speaker_id: 匿名化された話者識別子
  • duration: 音声の長さ(秒)
  • emotion: DeBERTa感情分析による8次元感情スコア
Downloads last month
5