Sobolev范数在神经嵌入中度量音频变形规律性
Sobolev Norms in Neural Embeddings Measure Audio Morphing Regularity
- Nantes Université, École Centrale Nantes, CNRS, LS2N, UMR6004（南特大学，南特中央理工学院，法国国家科学研究中心，LS2N，UMR6004）
- Arturia
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
本文提出Sobolev距离到理想变形（SDIM）作为客观度量，在感知相关的音频嵌入空间中量化变形轨迹的规律性，实验证明其优于现有度量，能可靠区分理想与对抗性轨迹。
AI中文摘要：
随着生成模型的出现，特别是在音频和图像生成领域，变形（morphing）最近重新引起了人们的兴趣。在音乐声音合成中，变形可以在两个目标之间生成中间声音，帮助音乐家和声音工程师探索具有有趣感知特性的新声音。由于变形本质上是在感知层面定义的，评估这一任务具有挑战性。在这项工作中，我们引入了Sobolev距离到理想变形（SDIM），一种新颖的客观度量，用于量化感知相关的音频嵌入空间中音频变形轨迹的规律性。利用基于物理的声音合成器，我们评估了SDIM在具有不同规律性程度的受控变形轨迹上的区分能力，并将其与现有音频变形度量进行了比较。结果表明，与最先进的度量相比，所提出的度量能够可靠地区分理想的轨迹与对抗性的轨迹。
英文摘要：
Morphing has recently gained renewed interest with the emergence of generative models, particularly in audio and image generation. In musical sound synthesis, morphing can generate intermediate sounds between two targets, helping musicians and sound engineers explore new sounds with interesting perceptual properties. As morphing is inherently defined in perceptual terms, evaluating this task is challenging. In this work, we introduce Sobolev Distances to Ideal Morphing (SDIM), a novel objective metric to quantify the regularity of audio morphing trajectories in perceptually relevant audio embedding spaces. Leveraging a physics-based sound synthesizer, we evaluate the discriminative power of SDIM on controlled morphing trajectories with varying degrees of regularity and compare it with that of existing audio morphing metrics. Results show that, contrary to state-of-the-art metrics, the proposed metric reliably discriminates desirable trajectories from adversarial ones.
