Randy Travis goes AI

By Chris Edwards

Last Friday, the legendary Randy Travis ended up a headline news item with a new song, “Where That Came From,” his first new recording since 2013, when he suffered a massive stroke.

There’s just one problem – sure, Travis was involved in the creation of the song, but the voice you hear is not actually Travis performing the tune in real time.

The stroke affected Travis in such a way that it left him paralyzed on his right side and damaged an area of his brain that controls speech and language, leaving him with limited speaking and singing ability. In an interview with his longtime producer, Kyle Lehning, the methodology behind the song was revealed: the vocal performance was an AI-generated version of Travis’s unmistakable, bourbon-smooth baritone.

I’m a huge fan of Randy Travis’s music. One of my fondest childhood memories linked to music involves a day trip to visit kinfolks in Sabine County, with Travis’s Storms of Life album playing in the tapedeck of my grandfather’s F-150.

To the general, casual-listening public, an AI-generated new song from Randy Travis might not make much difference, but as someone who loves the man and his music, I find this to be a huge slap in the face of the very authenticity that his music has always represented.

When Travis emerged from his near-death experience in 2013, about a year after it all happened, there wasn’t much in the press revealed about the state of his voice and/or any remaining disability from the stroke. In time, the extent was made public, and moments like him singing “Amazing Grace” at his Country Music Hall of Fame induction, in 2016, seemed like huge triumphs. The optimistic side of me hoped that with enough therapy and regaining of vocal skills, Travis could someday undergo a late-career renaissance, like what happened with Johnny Cash and his series of stark, mostly acoustic records in the 90s through the early aughts.

As explained in a CBS interview with Lehning, the song, written by Scotty Emerick and John Scott Sherrill, was co-produced by Lehning and Travis, and the way it went down goes like this: a singer, James Dupre, sang the song and an AI filter created from bits and pieces of songs Travis had sung throughout his career, was laid over Dupre’s performance for the end result.

Dupre is the vocalist who travels with Travis and his band, essentially doing a Randy Travis tribute act, although the great man is still very much alive. It makes sense that he would be involved with this project, and the Randy Travis AI filter is cobbled together from drips and drabs of actual Travis vocals. Still, there’s a lot that feels wrong about this project.

If Travis had a hand in writing the song, I might feel a little differently, maybe. Travis does, after all, have a disability. The aphasia that affects his speech limits what he can say, but all reports paint him as still fully in control, cognitively. Travis was always an incredible songwriter, and to many young musicians with aspirations of playing and singing songs, it probably never seemed fair that someone was gifted with a sound like that and could also write timeless songs, such as “I Told You So.”

Related Articles

Randy Travis goes AI

If Randy Travis were to write, or co-write, an album’s worth of songs and oversee its production, with guest vocalists taking on the singing duties, I’ll bet there’d be a phone ringing non-stop of monster talents begging to take part in such a project.

That premise, as I’m imagining it, brings me mind of an album that was released a few years ago, by the legendary neoclassical metal guitar shredder Jason Becker. Becker, who has been incapacitated by ALS for decades, painstakingly composed every note of music on the record and wrote lyrics for singers to sing, and he produced the project. Although Becker cannot even hold a guitar any longer, the album was purely his vision.

Such methods might bring up ages-old questions about “creating art by committee,” but when an artist is robbed of certain abilities to perform, it seems to be the next best thing, and absent the AI factor.

My question is this: why couldn’t Travis, with his limited vocal ability, sing a rough take of the song and then have his co-producer layer the AI filter over that?

In previous interviews with his wife Mary, it was revealed that although his speech is seriously compromised, singing comes much easier to Travis, after the stroke, but that, too, takes a lot of work. She stated that it was a meticulous process for him to be able to sing “Amazing Grace” at the aforementioned event.

For someone else to sing a song and have modern technology layer an effect (which is built from real-deal past Travis performances) seems, at the least, dishonest. The only consolation prize in this is knowing that Travis was involved, at least to a nominal degree.

The music of Randy Travis has always been built around authenticity; from his soulful, weathered-yet-smooth vocals to the subject matter of the songs he sung, Travis was never one to chase trends.

When this sort of computerized legerdemain is used to “bring back the voice” of a major artist like Randy Travis, the slope gets really slippery. Will the future of mainstream music be a world where artists no longer even have to perform or be involved on their recordings? In the end, from where I sit, it looks like AI could possibly spell the end of creative industries, and long-dead artists will suddenly “re-emerge” with “new” songs and albums – material they would’ve likely never even considered while they were here and able to sing for us.

AI in music, like in everything else, is a problem. It is not a good thing, and I’ll attest to that, forever and ever, amen.