Emotion-intensity Speech Synthesis on ESD & CNCED (Female)
4.28MOSEmoInstruct-TTS (Dual-Path)
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| EmoInstruct-TTS (Dual-Path)Evaluation mode=zero-shot, Variant=Dual-Path2026.06 | 4.28 | 4.25 | 4.52 | |
| CosyVoice2Evaluation mode=zero-shot2026.06 | 4.18 | 3.92 | 4.46 | |
| EmoInstruct-TTS -w/o Text Instruct (Emo2emb-Only)Evaluation mode=zero-shot, Variant=Emo2emb-Only2026.06 | 4.16 | 3.99 | 4.44 | |
| CosyVoice3Evaluation mode=zero-shot2026.06 | 4.15 | 3.98 | 4.47 | |
| EmoInstruct-TTS -w/o Emo2emb (Text-Only)Evaluation mode=zero-shot, Variant=Text-Only2026.06 | 4.12 | 3.78 | 4.4 |