Cross-style Speech Synthesis on DeepASMR-DB Cross-style Normal → ASMR
4.35ZH OI-MOSGround Truth
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Ground Truth2026.01 | 4.35 | 4.24 | 4.36 | 4.18 | |
| DeepASMRArchitecture=End-to-End2026.01 | 4.19 | 3.99 | 3.91 | 3.66 | |
| F5TTSVC Model=CosyVoiceVC2026.01 | 2.99 | 2.34 | 3.91 | 2.3 | |
| F5TTSVC Model=SeedVC2026.01 | 2.58 | 1.92 | 4.08 | 2.12 | |
| CosyVoice2VC Model=CosyVoiceVC2026.01 | 2.39 | 2.27 | 3.81 | 2.34 | |
| CosyVoice2VC Model=SeedVC2026.01 | 2.18 | 1.79 | 3.84 | 2.03 |