harsh13333/iemocap-dataset-instruct-2

该数据集包含了多种情感状态（如愤怒、悲伤、厌恶、兴奋、恐惧、中性、惊讶、快乐等），以及情感活动、情感价值和情感强度的评估。此外，数据集还包含了性别、转录文本、语速、音高平均值和标准差、均方根值、相对分贝等音频特征信息。视频信息也被包含在内，如视频文件名和摘要。数据集还记录了会话ID、对话ID、历史信息、对话指令以及对话消息的内容和角色。数据集分为训练集和测试集，共有8008个训练样本和2031个测试样本。

The dataset includes various emotional states such as anger, sadness, disgust, excitement, fear, neutral, surprise, happiness, as well as assessments of emotion activity, emotion value, and emotion dominance. In addition, the dataset contains audio features like gender, transcription text, speaking rate, mean and standard deviation of pitch, root mean square (RMS), relative decibels, and video information such as video filenames and summaries. The dataset also records session ID, dialogue ID, history, dialogue instructions, and the content and role of dialogue messages. The dataset is split into a training set and a test set, with a total of 8008 training examples and 2031 test examples.