Research Article
A Voice Cloning Method Based on the Improved HiFi-GAN Model
Table 5
Parameters, inference speed, MOS, and PESQ scores of different vocoders.
| Vocoder | MOS (CI) | PESQ | Parameters (M) | Speed on GPU | Speed on CPU |
| Ground truth | 4.56 0.08 | 4.48 | — | — | — | WaveNet | 3.97 0.06 | 3.35 | — | ×0.002 | — | WaveGlow | 3.96 0.07 | 3.19 | — | ×5.26 | ×0.13 | HiFi-GAN | 4.25 0.07 | 3.63 | 13.94 | ×70.34 | ×2.42 | Improved HiFi-GAN | 4.38 0.06 | 3.74 | 4.38 | ×78.67 | ×3.17 |
|
|