>Education, Technology, Social Sciences, Performing Arts, Computer Science

Deciphering the Secrets of Musical Intelligence: Big Data Analysis Techniques and End-User Tools for Music Score

Evaluating the feasibility of using sound to track honey bee recruitment dances

Deciphering the Secrets of Musical Intelligence:  Big Data Analysis Techniques and End-User Tools for Music Score
Musicians possess a wealth of knowledge, accumulated over the years of practice and experience of playing and studying music. For example, music educators are well aware of the characteristics of musical pieces that would make suitable repertoire for a given student. Certain parts of a music scores represent dissimilar levels of difficulty and can be played by beginner, intermediate, or advanced students. Information like that, referred to as musical knowledge, has been documented as guidelines used by musical education organizations. However, applying these guidelines to solve practical problems in musical performance and education remains hard, due to the sheer number of available scores and the difficulty of analyzing complex musical notation by hand. There is great potential benefit in rendering musical knowledge amenable for automated computer processing. Then powerful computer processing techniques can be developed to answer some of the most salient questions musicians have been asking daily. What pieces are suitable for this performing ensemble? What is the list of band scores that prominently feature a clarinet part? If my piano student has just successfully performed this piece, what should be their next one?

In this project, we have focused on the problem of rendering music knowledge amenable for automated computer processing and applying the result to different avenues. To that end, we have created the Toscanini API, a JavaScript framework that retrieves valuable information from written musical scores, such as key signature, pitch range, and rhythmic complexity. We are constantly enhancing and evolving the API to support new applications. To validate Toscanini, we have applied this API to two applications: GradeLevel and Translate. The GradeLevel application automatically grades individual scores, leveraging the manual guidelines provided through prior collective efforts of university music teachers. The Translate application automatically
translates musical notation into its natural language counterpart, thereby automatically describing musical scores in a way that non-musicians can understand. Our experiences with GradeLevel and Translate show the effectiveness of the Toscanini framework in automatically processing musical knowledge and opens up numerous additional opportunities to apply this technology.

Dr. Eli Tilevich, Department of Computer Science
Dr. Charles Nichols, School of Performing Arts

Collaborative Colleges:

Social Sciences
Performing Arts
Computer Science