Music Sampling Network Analysis (Python)

May 07, 2019 by Justin Tran

Summary poster of findings

What is musical sampling and why is it interesting?

Musical sampling influence has only recently been studied through network structures through the basic analysis of artist-artist sampling relationships. In this piece of research, we integrate the use of additional properties of music sampling (such as genre, time period, and audio element sampled) to investigate patterns of influence in the musical community at large.

How do we investigate musical sampling using graph theory?

Using the WhoSampled dataset and NetworkX for Python, we investigate statistical metrics such as the most-sampled artists songs as well as the trend for musical sampling over time. We also take a more nuanced look at “influence” by providing a variety of graph centrality measurements for determining the influence of a node (representing an artist) on other nodes.

What were the results of the research?

This analysis resulted in a greater understanding of musical influence certain artists and genres had over other heavily-sampling artists and genres over time. The most influential genre was found to be Funk/Soul/Disco while the most influential artist of all time was James Brown. However, Hip-Hop/R&B music has unsurprisingly become more sampled than all other genres in recent time periods. In addition, artists from all genres tended to sample from others within their own genre, essentially displaying strong intra-genre influences.

Where can I find the most pivotal details?

All code can be found in the Github repo.

The full paper can be found here.

The slide deck for a quick summary can be found here.

The high-resolution poster can be found here.