Speech Enhancement Based on Drifting Models
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866916029854121984 |
|---|---|
| author | Xu, Liang Caviedes-Nozal, Diego Kleijn, W. Bastiaan Yan, Longfei Felix Olsson, Rasmus Kongsgaard |
| author_facet | Xu, Liang Caviedes-Nozal, Diego Kleijn, W. Bastiaan Yan, Longfei Felix Olsson, Rasmus Kongsgaard |
| contents | We propose Speech Enhancement based on Drifting Models (DriftSE), a novel generative framework that formulates denoising as an equilibrium problem. Rather than relying on iterative sampling, DriftSE natively achieves one-step inference by evolving the pushforward distribution of a mapping function to directly match the clean speech distribution. This evolution is driven by a Drifting Field, a learned correction vector that guides samples toward the high-density regions of the clean distribution, which naturally facilitates training on unpaired data by matching distributions rather than paired samples. We investigate the framework under two formulations: a direct mapping from the noisy observation, and a stochastic conditional generative model from a Gaussian prior. Experiments on the VoiceBank-DEMAND benchmark demonstrate that DriftSE achieves high-fidelity enhancement in a single step, outperforming multi-step diffusion baselines and establishing a new paradigm for speech enhancement. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_24199 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Speech Enhancement Based on Drifting Models Xu, Liang Caviedes-Nozal, Diego Kleijn, W. Bastiaan Yan, Longfei Felix Olsson, Rasmus Kongsgaard Sound Artificial Intelligence Audio and Speech Processing Signal Processing We propose Speech Enhancement based on Drifting Models (DriftSE), a novel generative framework that formulates denoising as an equilibrium problem. Rather than relying on iterative sampling, DriftSE natively achieves one-step inference by evolving the pushforward distribution of a mapping function to directly match the clean speech distribution. This evolution is driven by a Drifting Field, a learned correction vector that guides samples toward the high-density regions of the clean distribution, which naturally facilitates training on unpaired data by matching distributions rather than paired samples. We investigate the framework under two formulations: a direct mapping from the noisy observation, and a stochastic conditional generative model from a Gaussian prior. Experiments on the VoiceBank-DEMAND benchmark demonstrate that DriftSE achieves high-fidelity enhancement in a single step, outperforming multi-step diffusion baselines and establishing a new paradigm for speech enhancement. |
| title | Speech Enhancement Based on Drifting Models |
| topic | Sound Artificial Intelligence Audio and Speech Processing Signal Processing |
| url | https://arxiv.org/abs/2604.24199 |