meldynamics/liepa2

该数据集包含音频文件及其相关标注信息，适用于语音识别、语音标注等相关研究领域。数据集分为训练集和测试集，音频采样率为16000Hz。每条数据包含音频ID、音频类型、路径、SHA1值、时长、标注文件路径、标签、分层信息、编解码器信息、说话风格、来源类型、性别、年龄组、说话人ID、标注员代码、会话ID、录音ID、层级说话人列表、清理后的转录文本、带噪声的转录文本、语音段落数量、噪声段落数量、语音总时长、噪声总时长、语音段信息、噪声段信息以及原始EAF XML内容等。

The dataset includes audio files and their corresponding annotation information, suitable for speech recognition, speech annotation, and other related research fields. The dataset is divided into training and test sets, with an audio sampling rate of 16000Hz. Each entry contains audio ID, audio type, path, SHA1 value, duration, annotation file path, tags, tier information, codec information, speech style, source type, gender, age group, speaker ID, annotator code, session ID, recording ID, tier speaker list, cleaned transcription text, transcription text with noise, number of speech segments, number of noise segments, total duration of speech, total duration of noise, speech segment information, noise segment information, and raw EAF XML content.