Speech Synthesis on LJSpeech
4.45MOSGround Truth
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Ground Truth2020.10 | 4.45 | — | — | |
| HiFi-GAN V1# Param (M)=13.922020.10 | 4.36 | 31.74 | 3,701 | |
| HiFi-GAN V2# Param (M)=0.922020.10 | 4.23 | 214.97 | 16,863 | |
| HiFi-GAN V3# Param (M)=1.462020.10 | 4.05 | 296.38 | 26,169 | |
| WaveNet (MoL)# Param (M)=24.732020.10 | 4.02 | — | 0.07 | |
| StyleTTS 22023.06 | 3.83 | — | — | |
| WaveGlow# Param (M)=87.732020.10 | 3.81 | 4.72 | 501 | |
| Ground Truth2023.06 | 3.81 | — | — | |
| MelGAN# Param (M)=4.262020.10 | 3.79 | 145.52 | 14,238 | |
| JETS2023.06 | 3.57 | — | — | |
| StyleTTS + HiFiGAN2023.06 | 3.35 | — | — | |
| VITS2023.06 | 3.34 | — | — |