Video-to-speech synthesis on LRS3
3.17MOS (Naturalness)HiCoDiT
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| HiCoDiT2026.04 | 3.17 | 2.88 | 3.5 | |
| Ground Truth2026.04 | 3.07 | 3.3 | 3.4 | |
| FTV2026.04 | 2.8 | 2.9 | 3.48 | |
| AlignDiT2026.04 | 2.47 | 2.63 | 3.13 |
| Method | Links | |||
|---|---|---|---|---|
| HiCoDiT2026.04 | 3.17 | 2.88 | 3.5 | |
| Ground Truth2026.04 | 3.07 | 3.3 | 3.4 | |
| FTV2026.04 | 2.8 | 2.9 | 3.48 | |
| AlignDiT2026.04 | 2.47 | 2.63 | 3.13 |