The evolution of musical trends over time has been a topic of growing interest, particularly in understanding how technological advancements and global events, such as the COVID-19 pandemic, influence the soundscape of popular music. Despite this interest, few studies have quantitatively profiled the extent of these temporal changes. This study presents a comprehensive analysis of the top 1000 songs on Spotify from 2011 to 2023, using a range of statistical and machine learning techniques to explore changes in musical features. Key musical attributes annotated by Spotify, as well as advanced audio features extracted using the Librosa library, such as Mel-Frequency Cepstral Coefficients (MFCCs), were analyzed using Principal Component Analysis (PCA), trend analysis (including linear regression and the Mann-Kendall test), Change Point Analysis, and hierarchical clustering. Results indicated significant shifts in musical characteristics, with pre-pandemic years marked by increasing trends in energy and loudness, whereas the post-pandemic period showed an increase in introspective attributes such as acousticness and instrumentalness. The years 2020 and 2021, coinciding with the COVID-19 pandemic, emerged as periods of notable change, characterized by distinct shifts in spectral and rhythmic attributes. Clustering analysis highlighted the emergence of a distinct group of softer, acoustic-driven songs during the pandemic. Our findings underscore the dynamic nature of popular music, driven by both technological advancements and external events. By providing a framework for analyzing temporal shifts in musical attributes, this study offers valuable insights for artists, producers, and industry stakeholders seeking to understand and anticipate evolving music trends.
Tan et al. (Fri,) studied this question.