Text-to-Speech on Emilia
2.2Word Error Rate (WER)WavCube-Pro
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| WavCube-ProNumber of Parameters=337M, TTS Training Data=95k h, zero-shot=true2026.05 | 2.2 | 70.9 | |
| Ground Truthzero-shot=true2026.05 | 2.23 | 69 | |
| F5-TTSNumber of Parameters=336M, TTS Training Data=95k h, zero-shot=true2026.05 | 2.42 | 66 | |
| FireRedTTSNumber of Parameters=580M, TTS Training Data=248k h, zero-shot=true2026.05 | 2.69 | 47 | |
| E2 TTSNumber of Parameters=333M, TTS Training Data=95k h, zero-shot=true2026.05 | 2.95 | 69 | |
| CosyVoiceNumber of Parameters=300M, TTS Training Data=170k h, zero-shot=true2026.05 | 3.59 | 66 |