Text-to-Speech on MiniMax multilingual zh, en, ja, ko (test)
75TPSFlashTTS
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| FlashTTSModel Configuration=MTP-5 (2-NFE)2026.06 | 75 | 62 | 206 | 123 | 328 | 0.621 | 20.8 | 0.668 | -0.08 | |
| FlashTTSModel Configuration=MTP-3 (2-NFE)2026.06 | 73 | 62 | 195 | 123 | 325 | 0.632 | 18.8 | 0.695 | 0.05 | |
| FlashTTSModel Configuration=MTP-3 (3-NFE)2026.06 | 72 | 62 | 215 | 165 | 366 | 0.702 | 17.5 | 0.714 | 0.12 | |
| CosyVoice2Model Configuration=10-NFE2026.06 | 51 | 257 | 425 | 339 | 843 | 0.913 | 26.2 | 0.721 | 0 | |
| FlashTTSModel Configuration=Stage 1 (2-NFE)2026.06 | 50 | 60 | 246 | 123 | 377 | 0.793 | 18 | 0.702 | 0.08 | |
| FlashTTSModel Configuration=Stage 1 (3-NFE)2026.06 | 50 | 60 | 246 | 198 | 413 | 0.824 | 17.2 | 0.711 | 0.15 |