The VIS Framework: Analyzing Counterpoint in Large Datasets
Christopher Antila and Julie Cumming
The VIS Framework for Music Analysis is a modular Python library designed for “big data” queries in symbolic musical data. Initially created as a tool for studying musical style change in counterpoint, we have built on the music21 and pandas libraries to provide the foundation for much more.
We describe the musicological needs that inspired the creation and growth of the VIS Framework, along with a survey of similar previous research. To demonstrate the effectiveness of our analytic approach and software, we present a sample query showing that the most commonly repeated contrapuntal patterns vary between three related style periods. We also emphasize our adaptation of typical n-gram-based research in music, our implementation strategy in VIS, and the flexibility of this approach for future researchers.
This paper was first presented at the 15th International Society for Music Information Retrieval (ISMIR) Conference, held between the 27–31 October 2014 in Taipei, Taiwan. For more information, refer to ismir2014.ismir.net.
Each test set comprises approximately fifty compositions, chosen to represent the time period in a balanced way. For convenience, each set is named after a representative composer from the period, though works by other composers are also present.