Stephen is working with musicians to design instruments that regenerate musical material into complementary sonic landscapes or respond with musical ideas of their own. He imagines an interactive space that enriches musical improvisation with sounds from nature or expansive arrangements that adapt to a musician's playing in real time.
Stephen Brade is an EECS PhD student at MIT. He uses live musical performance as a venue for research inquiry and creation, designing AI systems that take the stage alongside professional musicians. His work sits at the intersection of deep learning, human-computer interaction, and music technology, and is informed by his own practice as an Emerson-Harris Instrumental Jazz Scholar.
His ongoing research centers on building agentic AI systems that can meaningfully participate in live improvisation — responding to, challenging, and complementing human performers in real time. Through extended co-design processes with practicing jazz musicians, he develops new approaches to personalization, musical control, and the physical embodiment of AI within shared performance spaces. His research has been featured in public concerts at MIT, including "Agents in Concert," a performance showcasing three AI systems co-improvising with jazz musicians for a live audience.
His work has been published and presented at venues including ACM IUI and ISMIR. Prior to MIT, Stephen earned his BASc in Engineering Science and MSc in Computer Science from the University of Toronto, and interned at Adobe Research, where he conducted design-driven research on expressive text-to-speech for creative workflows.