[Paper Review] Cross-cultural Mood Perception in Pop Songs and its Alignment with Mood Detection Algorithms

Harin Lee, Frank Höger+3 more|arXiv (Cornell University)|Aug 2, 2021

Neuroscience and Music Perception39 references5 citations

TL;DR

This study investigates cross-cultural perception of mood in pop music across Brazil, South Korea, and the US, using a multilingual dataset of 360 songs rated for nine mood categories. Despite cultural differences in perception of complex moods like 'dreamy' and 'in love,' mood detection algorithms showed uniformly high alignment with human ratings across all cultures, suggesting their reliability as objective tools in popular music contexts.

ABSTRACT

Do people from different cultural backgrounds perceive the mood in music the same way? How closely do human ratings across different cultures approximate automatic mood detection algorithms that are often trained on corpora of predominantly Western popular music? Analyzing 166 participants responses from Brazil, South Korea, and the US, we examined the similarity between the ratings of nine categories of perceived moods in music and estimated their alignment with four popular mood detection algorithms. We created a dataset of 360 recent pop songs drawn from major music charts of the countries and constructed semantically identical mood descriptors across English, Korean, and Portuguese languages. Multiple participants from the three countries rated their familiarity, preference, and perceived moods for a given song. Ratings were highly similar within and across cultures for basic mood attributes such as sad, cheerful, and energetic. However, we found significant cross-cultural differences for more complex characteristics such as dreamy and love. To our surprise, the results of mood detection algorithms were uniformly correlated across human ratings from all three countries and did not show a detectable bias towards any particular culture. Our study thus suggests that the mood detection algorithms can be considered as an objective measure at least within the popular music context.

Motivation & Objective

• To examine whether people from different cultures perceive moods in pop music similarly.
• To assess the extent to which automated mood detection algorithms align with human ratings across diverse cultural groups.
• To investigate whether mood detection models trained on Western music exhibit cultural bias in cross-cultural settings.
• To create a multilingual, cross-cultural dataset of pop songs with semantically equivalent mood descriptors in English, Korean, and Portuguese.

Proposed method

• Collected 360 recent pop songs from national music charts in Brazil, South Korea, and the US.
• Developed semantically identical mood descriptors in English, Korean, and Portuguese to ensure cross-linguistic consistency.
• Gathered ratings from 166 participants across the three countries on familiarity, preference, and nine mood categories per song.
• Applied four state-of-the-art mood detection algorithms to the same song dataset for comparison with human ratings.
• Used correlation analysis to measure alignment between human ratings and algorithmic predictions across cultures.
• Conducted statistical tests to identify significant differences in perception across cultural groups for specific mood categories.

Experimental results

Research questions

• RQ1 Do individuals from Brazil, South Korea, and the US perceive the moods of pop songs similarly?
• RQ2 How well do automated mood detection algorithms align with human mood ratings across different cultural backgrounds?
• RQ3 Are mood detection algorithms biased toward Western cultural norms in their predictions?
• RQ4 Which mood attributes show the most significant cross-cultural variation in perception?

Key findings

• Human ratings of basic mood attributes such as 'sad', 'cheerful', and 'energetic' showed high similarity across all three cultures.
• Significant cross-cultural differences were observed for more complex mood descriptors such as 'dreamy' and 'in love', indicating cultural variation in perception.
• Mood detection algorithms demonstrated uniformly high correlation with human ratings across all three countries, with no detectable cultural bias.
• The algorithms' predictions were consistently aligned with human ratings regardless of the participants' cultural background.
• The study's multilingual dataset with semantically equivalent mood descriptors enabled valid cross-cultural comparisons.
• The results suggest that current mood detection algorithms can serve as objective, culture-robust tools for mood classification in popular music.