Can AI Restore a Lost Performance?

Dr. Bill Evans joins Real Music to break down a side of AI in music that most people never hear about: performance restoration. Instead of using AI to generate songs from scratch, Bill uses it to recover, reconstruct, and better reproduce the fine details of real performances by top-tier musicians. He explains why performance lives in the space between the notes on the page and the emotional impact a listener feels, and how that perspective shaped his work.

We get into how badly damaged live recordings from Flying Colors pushed him to develop what he calls “performance restoration,” including a case where seven seconds disappeared from the middle of a Steve Morse solo and he had to reconstruct what was played and what the intent of the performance was.

Bill also explains how his work evolved from early AI-assisted restoration into building virtual performance models, including a Marco Minnemann drum instrument that could preserve the exact feel of a recorded performance while allowing new MIDI-driven variations. From there, the conversation opens up into what it really means to restore, recreate, or even predict a performance at the artist level.

If you care about musicianship, production, AI, and where performance-level music technology is headed next, this is a fascinating one.