Reproduce “MUSEBENCH: A Benchmark for AI-Generated Music Detection and Artifact Analysis” 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
Yi Chen Liu, Syed Irfan Ali Meerza, Jiawei Yu, Xi Guo, Samuel Bright, Jian Liu
MUSEBENCH introduces a benchmark for music-authenticity detection covering human music, AI-generated tracks, human–AI remixes, and altered audio. It contains 361,166 tracks from ten generators. Detectors reach strong results on standard tests, but performance varies across unfamiliar or transformed music: AST achieves 100.0% balanced accuracy on the default test and 79.8% accuracy on a separate human-only out-of-domain set.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.