Paper

Absorbing Discrete Diffusion for Speech Enhancement

arXiv:2602.22417v2 Announce Type: replace-cross Abstract: Inspired by recent developments in neural speech coding and diffusion-based language modeling, we tackle speech enhancement by modeling the conditional distribution of clean speech codes given noisy speech codes using absorbing discrete diffusion. The proposed approach, which we call ADDSE, leverages both the expressive latent space of neural audio codecs and the non-autoregressive sampling procedure of diffusion models. To efficiently model the hierarchical structure of residual vector quantization codes, we propose RQDiT, which combi…

Authors:

Related discussions