Research Article
Emotion Modeling in Speech Signals: Discrete Wavelet Transform and Machine Learning Tools for Emotion Recognition System
Table 1
The average accuracy for each classifier across different emotion categories.
| Type | Sadness (%) | Neutral (%) | Happiness (%) | Disgust (%) | Boredom (%) | Anxiety-fear (%) | Anger (%) | Average (%) |
| SVM | 94.20 | 85.23 | 88.59 | 92.89 | 88.78 | 88.59 | 90.46 | 89.82 | KNN | 94.76 | 85.42 | 87.66 | 93.64 | 88.41 | 89.71 | 84.67 | 89.18 | Efficient Logistic Regression | 88.78 | 85.23 | 87.10 | 91.40 | 85.60 | 87.28 | 86.91 | 87.47 | Naive Bayes | 94.01 | 82.61 | 84.86 | 89.90 | 84.48 | 84.11 | 87.28 | 86.75 | Ensemble | 94.76 | 85.04 | 87.48 | 92.33 | 89.71 | 88.22 | 89.90 | 89.63 | Neural Network | 93.83 | 82.05 | 87.66 | 93.64 | 90.65 | 87.10 | 91.58 | 89.40 |
|
|
The “Average” column represents the average accuracy for each classifier.
|