On this week’s podcast episode, I speak with Dr. Bill Evans, audio scientist, producer, educator, and inventor of the Performance Restoration methodology.
Bill believes that drum recording has always involved a painful tradeoff, but it doesn’t have to.
Most engineers accept mic bleed as an unavoidable character of a drum sound, questioning whether removing it could actually alter the drummer’s true performance DNA.
Bill spent seven years answering exactly that question. His concept of the Virtual Audio Workstation, involves a new audio file format that remains compatible with WAV files while carrying additional performance metadata.
The format separates articulations, preserves room information as an independent element, and provides what Bill describes as functionally infinite dynamic range, allowing details like hi-hat foot checks to be raised dramatically without introducing grain or artifact.
At the center of that work is PRISM, Bill’s AI framework that converts audio to MIDI, edits it, and converts it back again, using the artist’s own recordings as the sole training data rather than a generative pool.
The distinction is important. Generative AI aims for plausible. PRISM aims for what a specific musician would actually play.
One of the earliest demonstrations came during a Flying Colors session when seven seconds of guitarist Steve Morse’s solo disappeared. Evans reconstructed the missing passage through PRISM, and Morse, known for being exacting about his performances, could not identify the recreated section.
The larger takeaway is a fundamental rethinking of how engineers approach drum recording.
By separating articulations and ambience into controllable elements, PRISM preserves ghost notes, cymbal strike positions, and subtle performance details that are typically lost in the production process, allowing the final recording to reflect the musician’s actual performance rather than a substitute version of it.
You can hear more at bobbyoinnercircle.com, or via Apple Podcasts, Amazon Music, YouTube Music, Mixcloud, Spotify, Deezer, TuneIn Radio, or RadioPublic.
Also, a video version of this podcast is now available on YouTube as well.
