Critically Auditing Generative Music Systems

My project examines the politics of AI systems that process, analyze, or produce sound, outlining the epistemologies and implications of voice recognition, signal processing, spatial analysis, and music generation. I use archival research, interviews with engineers and users, participant observation, analyses of scholarship and vernacular writing, and walkthroughs of technologies themselves. As my research partner, you will focus on music generation. My primary interest is for-profit corporate projects and start-ups. Please note that this is more an adversarial study of “unethical AI” and not a project in “ethical AI.” But artistic and open-source applications are also of interest.

I seek a research partner with strong programming skills for automated web applications and data analysis. Specifically, I will be asking you to build applications that “walk through” and audit different commercial AI tools for automated music generation that are available on the web, analyze the relationships between textual inputs and musical outputs, and compare across different platforms. Together, we will also come up with other creative methods to audit commercial generative music AI systems. A good working knowledge of machine learning, some background in the humanities/interpretive social sciences, strong writing skills, and an avid interest in music would be assets.