Reproduce “A Probabilistic Generative Model for Spectral Speech Enhancement” 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
Marco Hidalgo-Araya, Raphaël Trésor, Bart Van Erp, Wouter W.L. Nuijten, Thijs Van De Laar, Bert De Vries
SEM presents hearing-aid noise reduction as probabilistic inference, estimating filter settings from incoming sound rather than using fixed, hand-built processing steps. On the untrained VoiceBank+DEMAND test set, it reports BAK 3.47 and OVL 2.78, versus 2.68 and 2.67 for the time-frequency Wiener baseline, but lower PESQ, 2.17 versus 2.22. The study does not test adaptation to scenes or user preferences.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.