Emotional Text-to-Speech on CREMA-D
89.2Angry AccuracyDUET + GradTTS
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DUET + GradTTSPlug-and-play=true, Frozen backbones=true2026.05 | 89.2 | 23.1 | 68.2 | 60.2 | |
| DUET + MatchaPlug-and-play=true, Frozen backbones=true2026.05 | 61.8 | 10.8 | 69.8 | 47.4 | |
| DUET + ProDiffPlug-and-play=true, Frozen backbones=true2026.05 | 50.8 | 86.4 | 89.2 | 75.5 | |
| DUET + F5-TTSPlug-and-play=true, Frozen backbones=true, Reference speech with neutral emotion at inference=true2026.05 | 41.8 | 8.7 | 100 | 50.2 | |
| Qwen3-TTS2026.05 | 33.9 | 54.6 | 25.4 | 38 | |
| OpenAudio2026.05 | 33.9 | 39.5 | 10.3 | 27.9 | |
| EmoVoice2026.05 | 32.3 | 0 | 20.8 | 17.7 | |
| CosyVoice2Reference speech with neutral emotion at inference=true2026.05 | 28.5 | 31.3 | 39.2 | 33 | |
| DUET + StableTTSPlug-and-play=true, Frozen backbones=true2026.05 | 27.9 | 63.3 | 7.2 | 32.8 | |
| EmotiVoice2026.05 | 21.5 | 10.8 | 17.9 | 16.7 | |
| EmoKnobReference speech with neutral emotion at inference=true2026.05 | 7.4 | 42.6 | 59 | 36.3 | |
| Chatterbox2026.05 | 4.1 | 13.6 | 36.9 | 18.2 | |
| ChatTTS2026.05 | 0.2 | 23.6 | 26.4 | 16.7 | |
| IndexTTS22026.05 | 0 | 0 | 86.2 | 28.7 | |
| EmoSphere++2026.05 | 0 | 4.6 | 82.6 | 29.1 |