Abstract:

The study of cultural evolution through music has often focused on a single musical component (e.g., harmonic progression), overlooking how different subcomponents of music might independently interact within and across cultures. What hinders having such a comprehensive view is the limited cross-cultural dataset and the reliance on music transcription, which is labour-intensive and biased towards Western music. We introduce a novel methodological pipeline that bypasses the steps of transcription and works directly on the raw audio, enabling a large-scale, non-Western-centric study of global musical cultures. We compiled a novel dataset of 28,540 popular songs from 58 countries and analyzed their melodic and rhythmic regularities. Our approach empirically quantifies the global musical landscape and reveals distinct cultural characteristics that align with qualitative musicological research. Interestingly, we find melodic and rhythmic components demonstrate independent characteristics - a culture's variety in melody did not necessarily reflect variety in rhythm, nor does similarity to other cultures in one dimension correspond to similarity in the other. We found a strong correlation between a culture's ethnic diversity and rhythmic variability, but not melodic variability, suggesting rhythm may be more culturally ingrained and less permeable to change than melody.

Speaker's social media:

Twitter: @TweetHarin, BlueSky: @harinblue.bsky.social