Automatic Speech Recognition on Korean (ko) low-resource
7.2CERSylber 2.0
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Sylber 2.0Feature decomposition=Full2026.01 | 7.2 | 9.4 | |
| Sylber 2.0Feature decomposition=Content only2026.01 | 7.3 | 9.6 | |
| Mel2026.01 | 10.6 | 16 | |
| Mimi2026.01 | 20.9 | 31.6 | |
| Sylber2026.01 | 22 | 28.5 | |
| Sylber 2.0Feature decomposition=Acoustic only2026.01 | 22.3 | 32.4 | |
| DAC2026.01 | 100.3 | 125.3 |