发表机构
ETH Zurich; Sony Corporate Technology Center America, Inc.; Sony Europe Ltd.; Aalto University; Sony Group Corporation（苏黎世联邦理工学院; 索尼美国公司技术中心; 索尼欧洲有限公司; 阿尔托大学; 索尼集团公司）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出有监督的端到端管弦乐历史音乐修复基准，通过更忠实的合成退化链和潜流匹配模型，在多项评估中超越现有基线，并发布9.3小时测试集。
AI 中文摘要
历史音乐修复（HMR）几乎完全集中在受限问题上，如超分辨率或独奏作品的修复，而较少探索修复包含多种乐器的管弦乐历史音乐这一通用任务。这种探索不足主要是因为HMR领域，即20世纪初的录音，没有退化前的真实配对数据，使得修复任务成为无监督的，更具挑战性。本文通过探索合成退化函数和端到端生成式深度学习修复方法，提出了一个有监督的端到端管弦乐HMR基准。我们比先前工作更忠实地模拟了历史录音的退化链，使管弦乐修复成为一个可处理的有监督问题。在侵入式、非侵入式和主观评估中，基于所得合成配对训练的潜在流匹配模型优于现有的HMR基线。我们还策划并发布了一个9.3小时的无版权、未配对的历史古典音乐测试集，以及代码和音频演示。
英文摘要
Historical music restoration (HMR) has almost exclusively focused on constrained problems such as Super-Resolution or the restoration of solo pieces, under-exploring the general task of restoring orchestral historical music, which has multiple instruments. This under-exploration is largely because the HMR domain, early-20th-century recordings, has no pre-degradation ground-truth pairs, making the restoration task unsupervised and more challenging. This paper presents a supervised end-to-end orchestral HMR benchmark by exploring both the synthetic degradation functions and the end-to-end generative deep-learning restoration methods. We simulate the historical recording degradation chain more faithfully than prior work, which makes orchestral restoration into a tractable supervised problem. A latent flow-matching model trained on the resulting synthetic pairs outperforms existing HMR baselines on intrusive, non-intrusive, and subjective evaluations. We also curate and release a 9.3-hour license-free, unpaired, historical classical-music test set, along with code and audio demos.
Comments5 pages, 2 figures, 3 tables; submitted to ICASSP 2027. Code and audio demos available at the project repository
