2025 · Undergraduate R&D Scientist · Ontario Tech University
Generative AI for Live Opera
Meladramma-AI: GAN and diffusion models generating vocal textures and stage visuals for a live opera production
PythonGANsDiffusion modelsAudio ML
Meladramma-AI is a live opera production at Ontario Tech University that puts generative AI on the operatic stage. I built the AI that made it possible.
My role
I was the sole producer, coder, and manager of the AI production pipeline. I worked directly with Professor Carolyn (Bland) McGregor AM, Dean of the Faculty of Information Technology, and with the Faculty of Arts & Humanities to integrate AI into the production and elevate the audience experience. The work bridged a machine-learning workflow and a live performance under one hard, public deadline.
What I built
- AI vocal textures: generative models producing unique vocal timbres that performers and sound designers could weave into the score.
- Generated stage visuals: AI-generated visual elements projected during the performance, responding to the production’s themes.
- A production-grade pipeline: models that had to be reliable, art-directable, and fast enough for a live show, not just a notebook demo.
Why it matters
Most ML coursework ends at a validation metric. This work ended in front of a live audience, which meant owning the whole thing end to end, from the models to the moment the lights came up.