Wired’s piece about a North Carolina start-up named Zenph Sound Innovation teases out a lot of the nuances with respect to how advances in artificial intelligence are impacting casual music listening and academic music education:

Using complex software, North Carolina’s Zenph Sound Innovations models the musical performances of musicians from Thelonius Monk to Rachmaninoff, based on how they played in occasionally old, scratchy recordings. Using that model, the company creates new recordings as they would be played by deceased musicians, if they were around to record with today’s equipment, to critical acclaim. And that’s just for starters.

Zenph also plans explore a variety of new markets, including licensing clear versions of muddy recordings to films and software that could eventually let musicians jam with virtual versions of famous musicians.

Zenph’s first application of artificial musical personality is its specially designed robotic pianos, which take high-resolution MIDI files created by software that simulates the style of classical and jazz performers from days gone by, and turns them into sound by literally depressing the keys using between 12 and 24 high-resolution MIDI attributes. So far, the company offers new albums by legends including Art Tatum, Sergei Rachmaninoff and Glenn Gould, and up next is jazz pianist Oscar Peterson.