Armor
收藏arXiv2021-08-30 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/2108.12973v1
下载链接
链接失效反馈官方服务：
资源简介：
Armor是由北卡罗来纳大学教堂山分校创建的一个复杂且跨领域的基准数据集，旨在元评估人工智能音乐的质量。该数据集包含574个区分任务的音乐片段和248对（共496个）比较任务的音乐片段，涵盖21种不同音乐风格，用于推动人工智能音乐评估算法的发展。Armor数据集的设计考虑了音乐的多样性和复杂性，旨在解决人工智能音乐评估中的主观性和客观性之间的差距，为未来的人工智能音乐生成和计算音乐领域提供了一个全面且严格的评估框架。
Armor is a complex, cross-domain benchmark dataset created by the University of North Carolina at Chapel Hill for meta-evaluating the quality of artificial intelligence music. It contains 574 music clips for discrimination tasks and 248 pairs (totaling 496) of music clips for comparison tasks, covering 21 distinct music genres, and is designed to advance the development of AI music evaluation algorithms. Built with consideration of music diversity and complexity, the Armor dataset aims to bridge the gap between subjectivity and objectivity in artificial intelligence music evaluation, providing a comprehensive and rigorous evaluation framework for future AI music generation and computational music research fields.
提供机构：
北卡罗来纳大学教堂山分校
创建时间：
2021-08-30
