CAL10K (Computer Audition Lab 10000)

CAL10K 数据集（引入为 Swat10k）包含 10,870 首歌曲，这些歌曲使用 475 个声学标签和 153 个流派标签的标签词汇表进行了弱标记。这些标签都是从 Pandora 的网站上获取的，是由参与音乐基因组计划的专业音乐学家进行的歌曲注释的结果。The CAL10K dataset, also referred to as Swat10k, contains 10,870 songs that are weakly labeled using a label vocabulary encompassing 475 acoustic tags and 153 genre tags. All these tags were sourced from the Pandora website, and are the product of song annotations conducted by professional musicologists participating in the Music Genome Project.

搜集汇总

背景与挑战

背景概述

CAL10K数据集包含10,870首歌曲，通过475个声学标签和153个流派标签进行弱标记，这些标签源自Pandora网站并由专业音乐学家标注。该数据集由斯沃斯莫尔学院和Drexel University于2012年发布。