arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2411.02551cs.SDcs.AIcs.MMeess.AS

PIAST:一个包含音频、符号和文本的多模态钢琴数据集

PIAST: A Multimodal Piano Dataset with Audio, Symbolic and Text

  • Graduate School of Culture Technology, KAIST(韩国科学技术院文化技术研究生院)
  • NCSOFT(NCSoft公司)

机构由 AI 辅助整理,请以论文原文为准。

Hayeon Bang, Eunjin Choi, Megan Finch, Seungheon Doh, Seolhee Lee, Gyeong-Hoon Lee, Juhan Nam

更新

AI总结:

针对钢琴独奏音乐缺乏文本标签数据集的问题,构建了包含音频、符号和文本的多模态钢琴数据集PIAST,并验证了其在音乐标签与检索任务中的基线性能。

AI中文摘要:

虽然钢琴音乐已成为音乐信息检索(MIR)领域的重要研究方向,但目前明显缺乏带有文本标签的钢琴独奏音乐数据集。为解决这一空白,我们提出了PIAST(包含音频、符号和文本的钢琴数据集),一个钢琴音乐数据集。利用针对钢琴的语义标签分类体系,我们从YouTube收集了9,673首曲目,并由音乐专家为2,023首曲目添加了人工标注,从而形成两个子集:PIAST-YT和PIAST-AT。两个子集均包含音频、文本、标签注释以及利用最先进的钢琴转录和节拍跟踪模型转录的MIDI数据。在该多模态数据集的众多可能任务中,我们使用音频和MIDI数据进行了音乐标签分类和检索,并报告了基线性能,以展示其作为MIR研究宝贵资源的潜力。

英文摘要:

While piano music has become a significant area of study in Music Information Retrieval (MIR), there is a notable lack of datasets for piano solo music with text labels. To address this gap, we present PIAST (PIano dataset with Audio, Symbolic, and Text), a piano music dataset. Utilizing a piano-specific taxonomy of semantic tags, we collected 9,673 tracks from YouTube and added human annotations for 2,023 tracks by music experts, resulting in two subsets: PIAST-YT and PIAST-AT. Both include audio, text, tag annotations, and transcribed MIDI utilizing state-of-the-art piano transcription and beat tracking models. Among many possible tasks with the multi-modal dataset, we conduct music tagging and retrieval using both audio and MIDI data and report baseline performances to demonstrate its potential as a valuable resource for MIR research.

补充信息

↑