Instruction-based Text-to-Speech on UniSonate Instruction-based TTS Evaluation Set
100Emotion AccuracyGround Truth
Evaluation results
| Method | Links | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Ground Truth2026.04 | 100 | 100 | 100 | 100 | 100 | 100 | 1 | 1 | 0 | 0 | 0 | 0 | — | — | |
| InstructAudio2026.04 | 83.33 | 100 | 86.67 | 86.67 | 100 | 90 | 0.76 | 0.71 | 1.88 | 5.71 | 437.58 | 33 | 3.73 | 3.46 | |
| UniSonate2026.04 | 80 | 100 | 86.67 | 80 | 100 | 93.33 | 0.77 | 0.67 | 1.79 | 5.46 | 422.36 | 31 | 3.83 | 3.5 | |
| CosyVoice22026.04 | 58.33 | — | — | 65 | 100 | — | 0.68 | 0.53 | 2.57 | 7.11 | 547.87 | 46 | 3.9 | 3.65 |