Reproduce “Sobolev Norms in Neural Embeddings Measure Audio Morphing Regularity” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Théo Chasle Cauchy, Modan Tailleur, Barbara Pascal, Fanny Roche, Mathieu Lagrange
The authors introduce Sobolev Distances to Ideal Morphing (SDIM), a metric for comparing an audio trajectory’s whole path with a straight, evenly spaced reference in an embedding. Tested on controlled synthesized trajectories using MERT-v1-330M, SDIM distinguishes deliberately distorted paths that some existing metrics score close to ideal. This demonstrates discrimination in that setup, not prediction of human judgments.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.