Audio Customization on 70-example benchmark 1.0 (test)
59Speaker-SimFish-speech
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Fish-speechSettings=TTS, Background Sound=No2026.02 | 59 | 2.31 | |
| F5-TTSSettings=TTS, Background Sound=No2026.02 | 54 | 3.43 | |
| CosyVoiceSettings=TTS, Background Sound=No2026.02 | 53 | 4.58 | |
| OmniCustom (+ Contrastive learning losses)Settings=Sync Audio-video Customization, Background Sound=Yes2026.02 | 46 | 2.64 | |
| OmniCustom (+ Face&Timbre embeddings)Settings=Sync Audio-video Customization, Background Sound=Yes2026.02 | 38 | 2.82 | |
| OmniCustom (Baseline)Settings=Sync Audio-video Customization, Background Sound=Yes2026.02 | 27 | 2.37 |