Text-to-Speech on ZTTS1-Eval In-The-Wild (ja)
2.17WERFish S2 Pro
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Fish S2 Pro2026.06 | 2.17 | 2.73 | 75 | |
| Qwen 3 TTS 1.7B2026.06 | 2.32 | 2.86 | 75.2 | |
| Gemini 3.1 Flashsupports zero-shot voice cloning=false2026.06 | 2.72 | 3.56 | 28.3 | |
| Eleven Labs V3supports zero-shot voice cloning=false2026.06 | 2.84 | 3.49 | 28.6 | |
| Cartesia Sonic 3.52026.06 | 3.63 | 2.69 | 78.4 | |
| VoxCPM 22026.06 | 4.56 | 2.3 | 77.1 | |
| Inworld TTS 22026.06 | 6.93 | 2.81 | 63.3 | |
| ZONOS2 8B Quality ModeQuality Mode=true2026.06 | 8.7 | 2.69 | 70 | |
| Ground truth2026.06 | 9.18 | 2.27 | 82.3 | |
| ZONOS2 8BQuality Mode=false2026.06 | 10.18 | 2.34 | 70.9 |