Abstract: Investigating music with a focus on the similarity relations between songs, albums, and artists plays an important role when trying to understand trends in the history of music genres. In particular, representing these relations as a similarity network allows us to investigate the innovation presented by these entities in a multitude of points-of-view, including disruption. A disruptive object is one that creates a new stream of events, changing the traditional way of how a context usually works. The proper measurement of disruption remains as a task with large room for improvement, and these gaps are even more evident in the music domain, where the topic has not received much attention so far. This work builds on preliminary studies focused on the analysis of music disruption derived from metadata-based similarity networks, demonstrating that the raw audio can augment similarity information. We developed a case study based on a collection of a Brazilian local music tradition called Forró, that emphasizes the analytical and musicological potential of the musical disruption metric to describe and explain a genre trajectory over time.