MediaEval AcousticBrainz Genre

The AcousticBrainz Genre Dataset consists of four datasets of genre annotations and music features extracted from audio suited for evaluation of hierarchical multi-label genre classification systems. The datasets are used within the MediaEval AcousticBrainz Genre Task. The task is focused on content-based music<br> genre recognition using genre annotations from multiple sources and large-scale music features data available in the AcousticBrainz database. The goal of our task is to explore how the same music pieces can be annotated differently by different communities following different genre taxonomies, and how this should be addressed by content-based genre recognition systems. We provide four datasets containing genre and subgenre annotations extracted from four different online metadata sources: <strong>AllMusic</strong> and <strong>Discogs</strong> are based on editorial metadata databases maintained by music experts and enthusiasts. These sources contain explicit genre/subgenre annotations of music releases (albums) following a predefined genre namespace and taxonomy. We propagated release-level annotations to recordings (tracks) in AcousticBrainz to build the datasets. <strong>Lastfm</strong> and <strong>Tagtraum</strong> are based on collaborative music tagging platforms with large amounts of genre labels provided by their users for music recordings (tracks). We have automatically inferred a genre/subgenre taxonomy and annotations from these labels. For details on format and contents, please refer to the data webpage. Note, that the AllMusic ground-truth annotations are distributed separately at https://zenodo.org/record/2554044. <strong>Citation</strong> If you use the MediaEval AcousticBrainz Genre dataset or part of it, please cite our ISMIR 2019 overview paper: <pre><code>Bogdanov, D., Porter A., Schreiber H., Urbano J., &amp; Oramas S. (2019). The AcousticBrainz Genre Dataset: Multi-Source, Multi-Level, Multi-Label, and Large-Scale. 20th International Society for Music Information Retrieval Conference (ISMIR 2019).</code></pre> <strong>Acknowledgements</strong> This work is partially supported by the European Union’s Horizon 2020 research and innovation programme under grant agreement No 688382 AudioCommons.AcousticBrainz 音乐流派数据集（AcousticBrainz Genre Dataset）包含四类数据集，涵盖从音频中提取的流派标注与音乐特征，适用于分层多标签流派分类系统的评估。该数据集已应用于MediaEval AcousticBrainz流派任务（MediaEval AcousticBrainz Genre Task）中。本任务聚焦于利用多来源流派标注与AcousticBrainz数据库中大规模音乐特征数据开展基于内容的音乐流派识别。本次任务旨在探究两个核心问题：其一，不同社区依据不同流派分类体系，对同一音乐作品作出差异化标注的现状；其二，基于内容的流派识别系统应如何应对这一问题。 我们提供四类数据集，其流派与子流派标注均提取自四个不同的在线元数据源：AllMusic与Discogs均基于由音乐专家及爱好者维护的编辑式元数据库，二者均依据预定义的流派命名空间与分类体系，为音乐发行物（专辑）提供明确的流派/子流派标注。我们将专辑级别的标注迁移至AcousticBrainz中的录音（曲目），以此构建本数据集。Lastfm与Tagtraum则基于协作式音乐标签平台，该类平台汇集了大量用户为音乐录音（曲目）提供的流派标签。我们从这些标签中自动推导得到流派/子流派分类体系与标注信息。 关于数据集格式与内容的详细信息，请参阅数据官方页面。需注意，AllMusic的基准标注（ground-truth annotations）已通过https://zenodo.org/record/2554044 单独分发。 【引用说明】若您使用MediaEval AcousticBrainz流派数据集或其部分内容，请引用我们发表于第20届国际音乐信息检索大会（ISMIR 2019）的综述论文： Bogdanov, D., Porter A., Schreiber H., Urbano J., & Oramas S. (2019). The AcousticBrainz Genre Dataset: Multi-Source, Multi-Level, Multi-Label, and Large-Scale. 20th International Society for Music Information Retrieval Conference (ISMIR 2019). 【致谢】本研究部分受欧盟地平线2020研究与创新计划资助，项目编号为688382 AudioCommons。