SGMSE-BBED
This website contains supplementary material to the paper:
- Reducing the Prior Mismatch of Stochastic Differential Equations for Diffusion-based Speech Enhancement
The code is available at https://github.com/sp-uhh/sgmse-bbed
Denoising examples
| Filename | Clean | Noisy | OUVE (Baseline) [1] | BBED trs = 0.999 [2] | BBED trs = 0.5 [2] |
|---|---|---|---|---|---|
| 444c020o.wav | |||||
| 446c0210.wav | |||||
| 446c020h.wav | |||||
| 445c020j.wav | |||||
| 442o0301.wav | |||||
| 442c020d.wav | |||||
| 441c0207.wav | |||||
| 440o030e.wav | |||||
| 440c0207.wav |
References
[1] Julius Richter, Simon Welker, Jean-Marie Lemercier, Bunlong Lay, and Timo Gerkmann. Speech Enhancement and Dereverberation with Diffusion-Based Generative Models, EEE Trans. on Audio, Speech, and Language Proc. (TASLP), 2023
[2] Bunlong Lay, Simon Welker, Julius Richter, Timo Gerkmann "Reducing the Prior Mismatch of Stochastic Differential Equations for Diffusion-based Speech Enhancement", ISCA Interspeech, Dublin, Ireland, Aug. 2023