Research Article
Optimization of Spoken Term Detection System
Table 3
EER of different methods.
| Methods | EER |
| Baseline | 48.8% | CN | 47.5% | CN + maximum probability | 46.9% | MCE with sequential training | 46.2% | MCE with block training | 45.4% | CN + maximum probability + MCE with block training | 44.7% |
|
|