FlowSE: Flow matching-based speech enhancement
[video in korean]
[video in english]
Code here
Train set: WSJ0-CHiME3, Test set: WSJ0-CHiME3
| clean | noisy | Proposed Flow matching NFE=5 | SGMSE_CRP NFE = 5 [1] (reimplemented) | |
|---|---|---|---|---|
| 440c020a.wav |  |
 |
 |
 |
| 441o030v.wav |  |
 |
 |
 |
| 443o0313.wav |  |
 |
 |
 |
| 445o0303.wav |  |
 |
 |
 |
| 447o030s.wav |  |
 |
 |
 |
[1] Bunlong Lay, Jean-Marie Lermercier, Julius Richter and Timo Gerkmann. Single and Few-step Diffusion for Generative Speech Enhancement, ICASSP, 2024.