MUCaps

MUCaps Dataset

This is the MUCaps dataset, the largest music captioning dataset consisting of 21,966 music files with a total playtime of 1273.78 hours generated using the MU-LLaMA model.
This dataset is used to train the M2UGen model.
To uncompress the audio files, run the following:
cat mucaps_audios.tar.gz.* | tar xzvf -

The MUCapsCaptions.json file contains a dictionary with the filename as the key and the caption as the value.
This file is used to train the music encoder… See the full description on the dataset page: https://huggingface.co/datasets/M2UGen/MUCaps.

此乃MUCaps数据集，乃现存最大之音乐标题标注数据集，包含21,966个音乐文件，累计播放时长高达1273.78小时，由MU-LLaMA模型生成。本数据集旨在训练M2UGen模型。为解压音频文件，请执行以下命令：
cat mucaps_audios.tar.gz.* | tar xzvf -
MUCapsCaptions.文件中存储着一个字典，其中文件名为键，标题为值。此文件用于音乐编码器的训练……欲查阅数据集完整描述，请访问：https://huggingface.co/datasets/M2UGen/MUCaps。