Back to AIBriefs
AnalysisAI Models

Absorbing Discrete Diffusion for Speech Enhancement

Arxiv CS.SD (Sound/Audio ML)

Proposes an absorbing discrete diffusion method for speech enhancement. The approach models clean speech codes conditioned on noisy codes, inspired by neural speech coding and diffusion language models.

·
6 days ago
Absorbing Discrete Diffusion for Speech Enhancement — AIBriefs