Automatic Speech Recognition on Quechua (que) low-resource
30.1CERSylber 2.0
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Sylber 2.0Feature decomposition=Full2026.01 | 30.1 | 66.4 | |
| Mel2026.01 | 31.4 | 66.8 | |
| Sylber 2.0Feature decomposition=Content only2026.01 | 32.1 | 69.8 | |
| Mimi2026.01 | 39.9 | 76.8 | |
| Sylber2026.01 | 44.2 | 83.6 | |
| Sylber 2.0Feature decomposition=Acoustic only2026.01 | 48.8 | 86.1 | |
| DAC2026.01 | 51.1 | 87.3 |