Table 1 Predictive performance (concordance correlation coefficient) of the PVAN model and baselines for Valence and arousal.

From: Interpretable deep learning reveals distinct spectral and temporal drivers of perceived musical emotion

Model

Valence (CCC)

Arousal (CCC)

Hand-crafted Features

0.51

0.58

DeepSpectrum (CNN)

0.59

0.65

Music Transformer

0.62

0.68

PVAN (Full Model)

0.67

0.73

  1. Note: CCC = Concordance Correlation Coefficient. All models were evaluated on the DEAM dataset using 5-fold cross-validation.
Back to article page