Speech Synthesis on AISHELL3 Mandarin
2.7UTMOSGT
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| GT2026.03 | 2.7 | — | — | — | — | — | — | |
| BigVGAN‡Pre-trained=LibriTTS, Training Steps=5M steps2026.03 | 2.573 | 0.72 | 4.11 | 1.313 | 97 | 0.09 | 4.941 | |
| PeriodWave‡Pre-trained=LibriTTS2026.03 | 2.565 | 0.958 | 4.195 | 1.598 | 95.3 | 0.119 | 4.616 | |
| RNDVoC-sharedPre-trained=LibriTTS2026.03 | 2.486 | 0.702 | 4.049 | 1.267 | 96.8 | 0.091 | 4.912 | |
| HiFiGAN-V1Pre-trained=LibriTTS2026.03 | 2.432 | 1.059 | 2.823 | 2.827 | 93.4 | 0.163 | 4.651 | |
| BigVGAN-base‡Pre-trained=LibriTTS, Training Steps=5M steps2026.03 | 2.428 | 0.811 | 3.642 | 1.688 | 96 | 0.111 | 4.885 | |
| Vocos‡Pre-trained=LibriTTS2026.03 | 2.409 | 0.845 | 3.342 | 2.046 | 95.2 | 0.129 | 4.836 | |
| RNDVoC-nonsharedPre-trained=LibriTTS2026.03 | 2.402 | 0.752 | 3.938 | 1.445 | 96.8 | 0.093 | 4.886 | |
| iSTFTNet-V1Pre-trained=LibriTTS2026.03 | 2.351 | 1.132 | 2.595 | 3.061 | 93.2 | 0.164 | 4.486 | |
| PriorGradPre-trained=LibriTTS2026.03 | 2.197 | 1.054 | 3.995 | 1.717 | 91.7 | 0.16 | 4.675 | |
| FreGradPre-trained=LibriTTS2026.03 | 2.145 | 1.061 | 3.667 | 1.93 | 93.7 | 0.148 | 4.621 | |
| APNet2Pre-trained=LibriTTS2026.03 | 2.129 | 0.975 | 2.846 | 2.743 | 95 | 0.136 | 4.715 | |
| AvocodoPre-trained=LibriTTS2026.03 | 2.126 | 1.04 | 3.003 | 2.864 | 93.7 | 0.154 | 4.731 | |
| WaveFM‡Pre-trained=LibriTTS2026.03 | 2.037 | 0.832 | 3.81 | 1.599 | 96.4 | 0.108 | 4.929 |