Text-to-Speech on ZTTS In-The-Wild pl 1 (Eval)
4.66WEREleven Labs V3
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Eleven Labs V3supports zero-shot voice cloning=false2026.06 | 4.66 | 3.37 | 23.6 | |
| Cartesia Sonic 3.52026.06 | 5.3 | 2.73 | 79.5 | |
| Gemini 3.1 Flashsupports zero-shot voice cloning=false2026.06 | 5.33 | 3.47 | 26.2 | |
| ZONOS2 8B Quality ModeQuality Mode=true2026.06 | 5.85 | 2.76 | 69 | |
| VoxCPM 22026.06 | 7.28 | 2.35 | 78 | |
| Inworld TTS 22026.06 | 7.65 | 2.8 | 60.6 | |
| Fish S2 Pro2026.06 | 8.25 | 2.67 | 73.5 | |
| ZONOS2 8BQuality Mode=false2026.06 | 10.16 | 2.48 | 72.9 | |
| Ground truth2026.06 | 14.1 | 2.41 | 82.5 | |
| Qwen 3 TTS 1.7B2026.06 | 82.04 | 2.92 | 62.7 |