Reproduce “AudioFM: Towards General Purpose Audio Representation Learning” 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

Yamin Li, Ruidong Zhang, Anran Wang, Eduardo Fonseca, Dan Ellis, Zechen Li, Simon A. Lee, Tao Pang and 14 more

AudioFM learns audio representations by aligning masked recording views and predicting masked patches from a full-view teacher. Its objective retains event information alongside recording-level context; a one-step temporal neighborhood outperforms center-only supervision in the tested setup. AudioFM ranks best or tied on 10 of 12 classification tasks and reaches 79.9% macro mAP on 12 mapped RealDESED classes without target-domain training.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.