Figure 1.
A bipartite network that connects music styles with instruments is constructed.
(A) Schematic representation of the data containing the relations between styles and instruments. (B) Visualization of the matrix describing the music production network, . A black (white) field for style
and instrument
indicates that
. (C) Part of the bipartite network
that connects music styles with instruments for a given year
. Large nodes represent music styles, small ones instruments. It is apparent that some instruments occur in almost every style while others are used by a substantially smaller number of styles. For instance, there are only two instruments appearing exclusively in ‘hip hop’ among these five styles (for example the flageolet) whereas dozens of instruments are only related to ‘experimental’ (such as countertenor vocals). Vocals, lead guitar, and drums, on the other hand, appear in each of the five styles, whereas bones used as percussion elements only appear in ‘Black Metal’.
Figure 2.
Maximum spanning tree for the style-similarity network, , for the years 2004–2010.
Nodes represent styles, colors correspond to the genre to which the style belongs, the node size is proportional to the number of albums released for each style, the link strength between two music styles and
is proportional to
. Several clusters are visible. They are identified as styles belonging to ‘rock’, ‘jazz’, or ‘electronic music’ genres.
Figure 3.
Instrumentational variety and uniformity
for music styles within the time period t = 2004–2010.
Music styles collapse onto a line where and
are inversely related. ‘Experimental’ is the music style with the highest instrumentational variety, styles with the lowest levels of variety and highest uniformity values belong to the ‘electronic’ and ‘hip hop’ genres. Inset: The values for
and
are similar to results from the model.
Figure 4.
The arrangement of styles in the V-U plane remains robust over more than fifty years of music history.
However, the position of individual styles can change dramatically over time, as it is shown for ‘indie rock’, ‘new wave’, ‘disco’ and ‘synth-pop’. Some styles, such as ‘folk’, show almost no change in their position.
Figure 5.
Changes in instrumentational complexity of a style are related to its number of sales and to the number of artists contributing to that style.
(A) Changes in number of albums versus changes in instrumentational complexity of music styles
. We find a positive trend between
and
with correlation coefficient
and p-value
. (B) Sales
show a negative trend when compared to the change of complexity of music styles with correlation coefficient
and p-value
.