Medley2K

Medley2K是由苏黎世联邦理工学院创建的一个包含2000个由不同歌曲通过平滑过渡连接而成的音乐混搭数据集。该数据集包含7712个标记的过渡点，涵盖多种音乐风格和过渡类型。数据集的创建过程涉及从musescore.com网站爬取人类创作的混搭作品，并使用机器可读的MXL文件进行过渡点的精确标注。Medley2K旨在支持自动音乐混搭生成领域的研究，特别是解决音乐过渡生成的挑战。

Medley2K is a dataset consisting of 2000 music medleys, each formed by seamlessly connecting distinct songs, created by ETH Zurich. This dataset contains 7,712 annotated transition points, covering a wide range of music genres and transition types. The construction of this dataset involved crawling human-created medley works from the musescore.com website, and precisely annotating transition points using machine-readable MXL files. Medley2K aims to support research in the field of automatic music medley generation, particularly addressing the challenges of music transition generation.