Recomposing Data: Machine Learning as Compositional Process

This exposition reflects on how machine learning can be integrated with algorithmic composition and live coding to expand digital music creation. The research examines how ML-driven sound analysis, training data, and interactive models reshape compositional workflows. By viewing machine learning as an interpretative and generative process rather than a mere tool, this project challenges conventional boundaries between data gathering, system design, and artistic practice. The discussion is framed through experimental approaches that merge sound synthesis, live coding, and model training, questioning how algorithmic systems can act as both agents of composition and reflective mirrors of musical intention. Through the interplay of structured data, generative models, and exploratory workflows, the study situates machine learning within a broader conversation about creativity, computation, and the evolving role of the composer-programmer. keywords: machine learning, live coding, sound synthesis, KonCon Lectorate