Audio Generation on LJSpeech Short-Term (test)
0.911FADBemaGANv2
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| BemaGANv2Discriminator Configuration=MED + MRD2025.06 | 0.911 | 0.85 | 0.949 | 1.652 | 1.262 | 0.161 | |
| BemaGANv2Discriminator Configuration=MED only2025.06 | 1.116 | 0.82 | 0.94 | 1.887 | 1.453 | 0.2208 | |
| HiFi-GANComponents=AMP + Snake2025.06 | 1.5363 | 0.78 | 0.888 | 2.553 | 1.527 | 0.1543 | |
| BigVGANDiscriminator Configuration=MPD + MRD2025.06 | 1.906 | 0.79 | 0.906 | 2.56 | 1.519 | 0.1716 | |
| MED + MPD + MRD2025.06 | 2.146 | 0.71 | 0.851 | 3.136 | 1.771 | 0.1692 | |
| MSD + MED2025.06 | 2.337 | 0.67 | 0.879 | 2.964 | 1.803 | 0.1735 | |
| BemaGANDiscriminator Configuration=MPD + MED2025.06 | 3.194 | 0.68 | 0.861 | 3.18 | 1.893 | 0.1883 | |
| MSD + MRD2025.06 | 3.327 | 0.63 | 0.841 | 3.354 | 1.948 | 0.1802 | |
| HiFi-GANDiscriminator Configuration=MPD + MSD2025.06 | 6.06 | 0.73 | 0.792 | 3.22 | 3.256 | 0.1883 |