Decoding Musical Evolution Through Network Science

TL;DR

This study applies network science to analyze musical evolution by representing MIDI-based compositions as weighted directed networks (notes as nodes, transitions as edges) drawn from ≈$20{,}000$ files across $6$ macro-genres over nearly $4$ centuries. Topological metrics (e.g., density, reciprocity, entropy, global efficiency) characterize melodic complexity, and two embedding schemes— a $12$-dimensional interval vector $v_G$ and graph2vec embeddings— map networks into high-dimensional spaces, enabling genre clustering via $UMAP$. A release-date estimation pipeline combining a Large Language Model Gemini and Spotify data enables temporal trends, with formal tests (e.g., Mann–Kendall) confirming a long-run simplification in Classical and Jazz while other genres remain relatively flat. The results suggest that digital democratization and platform-driven connectivity foster the rise of simpler, more homogeneous genres, even as legacy genres retain higher intrinsic complexity; the framework offers a quantitative bridge between musicology, network science, and digital culture, though limitations in genre tagging and metadata remain.

Abstract

Music has always been central to human culture, reflecting and shaping traditions, emotions, and societal changes. Technological advancements have transformed how music is created and consumed, influencing tastes and the music itself. In this study, we use Network Science to analyze musical complexity. Drawing on $\approx20,000$ MIDI files across six macro-genres spanning nearly four centuries, we represent each composition as a weighted directed network to study its structural properties. Our results show that Classical and Jazz compositions have higher complexity and melodic diversity than recently developed genres. However, a temporal analysis reveals a trend toward simplification, with even Classical and Jazz nearing the complexity levels of modern genres. This study highlights how digital tools and streaming platforms shape musical evolution, fostering new genres while driving homogenization and simplicity.

Figures (16)

• Figure 1: $(a)$ CCDF of aggregate network's weights. $(b)$ distribution of various measures, divided according to the genre. The vertical bars represent the quartiles of each distribution.
• Figure 2: $(a)$ Distribution of unweighted efficiency. $(b)$ Distribution of weighted efficiency. $(c),(d)$ comparison between efficiency and weighted efficiency in real and randomized networks. To improve their visualization we use a random sample of $n = 10^4$ points.
• Figure 3: $(a)$ Fraction of intervals appearing in each musical genre. $(b)$ Center of mass for each genre, computed using UMAP $2-$dimensional coordinates. $(c)$ correlation between components and measures. $(d)$ correlation between intervals entries and components.
• Figure 4: $(a)$ Evolution of mean efficiency measures over decades. The arrows highlight the temporal evolution of considered eras. $(b)$ Distribution of genres in each musical era. $(c)$ Center of mass of each musical period, obtained using UMAP on the interval embeddings.
• Figure 5: Networks constructed from four MIDI files. The size of each node is proportional to the degree of the node and the transparency of each edge is proportional to its weight.