Research Article
Reordering Features with Weights Fusion in Multiclass and Multiple-Kernel Speech Emotion Recognition
Table 3
Emotion recognition accuracies of different feature selection methods with the best feature number.
| Selection methods | Feature number | SER accuracies (%) | Total | Angry | Fear | Happy | Neutral | Sad |
| DISR | 39 | 88.614 | 93.651 | 79.412 | 97.143 | 74.359 | 96.774 | MRMR | 39 | 90.099 | 92.063 | 67.647 | 97.143 | 94.872 | 96.774 | RFWF MK | 15 | 90.594 | 90.476 | 79.412 | 97.143 | 87.179 | 100 | RFWF SK | 39 | 79.208 | 95.238 | 73.529 | 28.571 | 87.179 | 100 |
|
|