Qualitative Inquiry on Machine Learning in Musical Performance
This project studies how machine learning becomes part of musical performance practice, and how artists develop techniques, shared knowledge and critical relationships with these systems.
Overview
As in many artistic sectors, Machine Learning (ML) has become part of music performance practice, although it remains little studied. Such inquiry can provide important insights into modes of expression and interaction, and into the collective practices of artistic communities.
We conducted an interview study with 14 musical artists about their relationship with ML. We first found that artists developed new interaction strategies with ML to enable musical agency: familiarizing themselves with the technology, controlling its behavior and exploring its limits in live performances.
These strategies are developed through data curation, real-time interaction and long-term practice. We also found that artists have a practice of remixing and assembling musical material that extends to the collective level through the sharing of knowledge and content.