Our lab aims to intersect these two domains by creating interfaces that are intuitive and highly conducive to musical expression. These interfaces are used to control parameters within generative AI algorithms, enabling a seamless blend of human creativity and machine-generated music. Our vision is to offer a spectrum of expression: from triggering and controlling individual notes to navigating the general features of a piece through gestures and other sensor-based inputs. We develop these interfaces with Arabic music primarily in mind, which has distinct features compared to Western music, such as unique scales, rhythms, and nonstandard tuning systems.

Generative AI is taking the world by storm, and music is no exception. Recent models are capable of generating realistic music by learning from text-to-audio data, opening up new horizons for musical creation. This development poses intriguing questions: what does this mean for musicians and the way we create music in the future?

Our lab explores these questions in the context of various tuning systems, focusing on real-time generation for improvisation. We aim to understand how generative AI can be integrated into musical practices to enhance creativity while expanding upon the Arabic music tradition with its unique scales and non-standard tuning systems.

Machine learning can uncover patterns in large datasets, and reveal insights that may not be apparent through human analysis alone. In this project, we develop tools to enhance the musicological understanding of Arabic music within its cultural, geographical, and historical contexts. We focus on music information retrieval (MIR) algorithms that address the unique challenges of Arabic music, such as accounting for quarter tones and complex modal structures. We aim to shed new light on the rich traditions of Arabic music and facilitate further research that is data-driven rather than starting from assumed theoretical frameworks.

Collaborator:
Dr. Dany Abou Jaoude