📊 Showing 6 results | 📏 Metric: PESQ-WB
| Rank | Model | Paper | PESQ-WB | Date | Code |
|---|---|---|---|---|---|
| 1 | Schrödinger Bridge (PESQ loss) | Investigating Training Objectives for Generative Speech Enhancement | 0.73 | 2024-09-16 | 📦 sp-uhh/sgmse |
| 2 | SGMSE+ | Speech Enhancement and Dereverberation with Diffusion-based Generative Models | 0.73 | 2022-08-11 | 📦 sp-uhh/sgmse |
| 3 | Schrödinger Bridge | Schrödinger Bridge for Generative Speech Enhancement | 0.73 | 2024-07-22 | - |
| 4 | Demucs v4 | Hybrid Transformers for Music Source Separation | 0.71 | 2022-11-15 | 📦 facebookresearch/demucs 📦 zhaozhipeng1997/demucs_ascend910_pytorch |
| 5 | Conv-TasNet | Conv-TasNet: Surpassing Ideal Time-Frequency Magnitude Masking for Speech Separation | 0.70 | 2018-09-20 | 📦 facebookresearch/demucs 📦 mpariente/asteroid 📦 facebookresearch/svoice |
| 6 | CDiffuSE | Conditional Diffusion Probabilistic Model for Speech Enhancement | 0.53 | 2022-02-10 | 📦 neillu23/cdiffuse 📦 christopher-beckham/annotated-conditional-diffusion |