Speech Synthesis on MiniMax multilingual Japanese (test)
3.52Word Error Rate (WER)MiniMax
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MiniMaxCommercial Model=true2026.06 | 3.52 | 77.6 | |
| CosyVoice2Commercial Model=false2026.06 | 7.94 | 80.3 | |
| FlashTTSCommercial Model=false2026.06 | 10.59 | 75.1 | |
| ElevenLabsCommercial Model=true2026.06 | 10.65 | 73.8 |