Table 2 Accuracy (in %) of the audio segmentation system with different number of HMM states and component Gaussians
From: Automatic segmentation of infant cry signals using hidden Markov models
No. of | No. of HMM states for each class | |||
|---|---|---|---|---|
Gaussians | 1 | 2 | 3 | 4 |
5 | 85.0 ±1.3 | 87.1 ±1.2 | 87.8 ±1.2 | 87.7 ±1.1 |
10 | 85.7 ±1.2 | 85.4 ± 1.1 | 87.5 ±1.1 | 87.6 ± 1.1 |
15 | 85.7 ± 1.2 | 85.5 ± 1.0 | 87.9 ±1.1 | 87.9 ±1.0 |
20 | 86.1 ±1.1 | 87.8 ± 1.0 | 87.8 ±1.0 | 88.1 ±1.1 |
25 | 86.1 ±1.1 | 88.0 ±1.0 | 87.9 ±1.0 | 88.2 ±1.1 |