kelvincai/MusicCaps_30s

该数据集是一个包含音频文件的集合，每个音频文件都有对应的元数据，如ytid（视频ID）、时间戳（start_s和end_s）、音频标签（audioset_positive_labels）、属性列表（aspect_list）、字幕（caption）、作者ID（author_id）等。数据集还包括是否为平衡子集的标识（is_balanced_subset）和是否为Audioset评估集的标识（is_audioset_eval）。音频的采样率为44100Hz。数据集提供了一个训练集split，其大小为9521420498.655653字节，包含1710个示例。整个数据集的下载大小为9167042480字节。

This dataset is a collection of audio files, each with associated metadata such as ytid (video ID), timestamps (start_s and end_s), audio labels (audioset_positive_labels), aspect lists (aspect_list), captions (caption), author IDs (author_id), etc. The dataset also includes flags for whether it is a balanced subset (is_balanced_subset) and whether it is an Audioset evaluation set (is_audioset_eval). The audio sampling rate is 44100Hz. The dataset provides a training set split, which is 9521420498.655653 bytes in size and contains 1710 examples. The total download size of the dataset is 9167042480 bytes.