From: Comparative study of singing voice detection based on deep neural networks and ensemble learning
Stacking method | Jamendo dataset | FMA dataset (%) |
---|---|---|
Voting, 5 models | 94.2% | 88.9 |
Voting, 3 models | 93.8 | 88.6 |
Post classifier, 5 model, approach 1 | 94.3% | 88.7 |
Post classifier, 3 model, approach 1 | 93.8% | 88.9 |
Post classifier, 5 model, approach 2 | 94.3% | 89.1 |
Post classifier, 3 model, approach 2 | 93.0% | 88.5 |