talkpl-ai/TalkPlayData-2

TalkPlayData 2是一个用于会话式音乐推荐的人工合成数据集，具有多模态（音频、专辑艺术、歌词、元数据）的特点，由代理框架创建。该数据集用于训练和评估TalkPlay-Tools系统中的会话式音乐推荐。数据集包括会话的目标、用户资料、对话内容等信息，并提供了训练集和测试集的统计数据和用户人口统计分布。数据集的创建和用途在README中进行了详细描述。

TalkPlayData 2 is a synthetic dataset for conversational music recommendation, which is multimodal (audio, album art, lyrics, metadata) and created using an agentic framework. This dataset is used for training and evaluation in the TalkPlay-Tools system. It includes information about conversation goals, user profiles, conversation content, and more. The README provides statistics for the training and test sets as well as user demographic distributions. The creation and usage of the dataset are described in detail in the README.