Abstract
National anthems serve as powerful symbols of national identity, often evoking strong emotional responses. While prior research has examined anthem lyrics, the emotional content of their musical features remains underexplored. This study employs computational modeling to predict the perceived emotional characteristics of 176 national anthems and investigates geographical and cultural variations. Using perceptual data from a prior study and musical features extracted with the MIR Toolbox, we trained LASSO regression models to predict eight emotional characteristics: Valence, Energy Arousal, Tension Arousal, Happiness, Sadness, Tenderness, Anger, and Fear. The predicted emotions were analyzed for continental differences, correlated with latitude and longitude, and compared to Hofstede's cultural dimensions. The results revealed significant geographic trends, with Valence lower in the Americas and Energy Arousal higher near the equator. Fear and Tension Arousal were more pronounced in the Americas, while Happiness was highest in Oceania. Cultural analyses indicated that hierarchical societies exhibited more energetic anthems, individualistic cultures had less tense but more tender anthems, and indulgent societies expressed greater Fear. These findings highlight the role of musical features in shaping anthem emotions and underscore the potential of computational approaches for large-scale music-emotion research.