Table 2 F1 Scores in frames, frame error rates and phone error rates for each model
From: Theoretical learning guarantees applied to acoustic modeling
| Classifier | F1 Score | FER% | PER% |
|---|---|---|---|
| GMM-HMM | 0.166 | 76.36 | 75.17 |
| CNN-MLP | 0.225 | 56.97 | 52.90 |
| CNN-HTSVM | 0.491 | 37.04 | 35.41 |