ASMR-to-Normal (A2N) Style Synthesis on ASMR Cross-Style Synthesis
33.8Global Unvoiced RatioGroundTruth
Evaluation results
| Method | Links | |
|---|---|---|
| GroundTruth2026.01 | 33.8 | |
| DeepASMRTTS Model=DeepASMR2026.01 | 39.99 | |
| CosyVoice2 + SeedVCTTS Model=CosyVoice2, VC Model=SeedVC2026.01 | 56.63 | |
| CosyVoice2 + CosyVoiceVCTTS Model=CosyVoice2, VC Model=CosyVoiceVC2026.01 | 57.23 | |
| F5TTS + CosyVoiceVCTTS Model=F5TTS, VC Model=CosyVoiceVC2026.01 | 58.84 | |
| F5TTS + SeedVCTTS Model=F5TTS, VC Model=SeedVC2026.01 | 65.93 |