Reproduce “SNOAR: Self-training with Noise filtering, Observation adding, and Remixing for Speech Denoising Without Clean References” 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

Samuele Cornell, Matthew Maciejewski

SNOAR trains speech denoisers without clean training references by filtering estimated noise from a teacher’s speech estimate, then blending some original noisy audio back in before remixing targets for a student. On WHAM!, it reached 13.25 dB SI-SDR, below the supervised 13.68 dB. On tested CHiME-3 channel-five recordings, it improved all reported metrics for three teachers.

Summary by alphaXiv.

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