Suno70k
收藏资源简介:
Suno70k是由中国科学院团队构建的大规模开源AI歌曲数据集,包含7万条高质量音乐样本,配备增强标签与歌词注释。该数据集通过系统化采集和标注流程构建,旨在解决全曲生成领域缺乏开源高质量数据的问题,为覆盖歌曲生成、音乐风格迁移等研究提供基础支持。其多维度标注特性特别适用于需要旋律控制与歌词同步的生成任务,推动了AI音乐创作的边界扩展。
Suno70k is a large-scale open-source AI song dataset constructed by a team from the Chinese Academy of Sciences. It contains 70,000 high-quality music samples, equipped with enhanced tags and lyric annotations. Built through a systematic collection and annotation pipeline, this dataset aims to address the shortage of open-source high-quality data in the field of full-song generation, providing foundational support for research covering song generation, music style transfer, and other related domains. Its multi-dimensional annotation features are particularly suitable for generation tasks that require melody control and lyric synchronization, promoting the expansion of the boundaries of AI music creation.
- 1SongEcho: Towards Cover Song Generation via Instance-Adaptive Element-wise Linear Modulation中国科学院·自动化研究所; 中国科学院大学·人工智能学院; 中国科学院·软件研究所; 康斯坦茨大学; 国立成功大学 · 2026年



