Table 5.
Comparison between performance of models (in terms of % accuracy) implemented on RAVDESS dataset
| SVM | Decision tree | CNN | |
|---|---|---|---|
| Model accuracy on original data | 30.17 | 77 | 75 |
| Model accuracy after applying autoencoder | 40.16 | 76 | 80 |
| Average speedup in accuracy (%) | 33.11 | − 1.29 | 6.66 |