TIOGA, Texas — In the world of country music, few names evoke the kind of reverence that Randy Travis’ does.
His voice — rich, soulful, unmistakable — defined a generation of performers and became synonymous with classic country music.
But for over a decade, after a stroke that left him unable to sing, Travis’ fans had been deprived of hearing that voice.
“For 11 years, I wanted to hear his voice — and I never thought I would get to hear it again,” said Travis’ wife Mary.
But now, with the help of artificial intelligence, there’s a new chapter in his remarkable story.
It started as a question posed by record executives at his Warner Bros. label.
“What if we can give Randy his voice back?” Travis’ longtime producer Kyle Lehning recalled being asked in a phone conversation with a label executive one day.
His response wasn’t immediately positive — not upon initially hearing the details behind her pitch, anyway.
“When she said artificial intelligence, my immediate response was, ‘That’s creepy,'” Lehning said.
For years, Travis’ fans held on to the hope that he would recover well enough to perform again. But the massive stroke left Travis with aphasia, a condition that impairs the ability to communicate, and robbed him of the ability to sing.
So, in late 2023 Lehning worked with the help of a London-based tech firm to build a unique AI model of Travis’ voice using his catalog of recordings from 1985 to 2010. That model would help them make Travis’ first song in almost a decade — a song called, “Where That Came From.”
“[They took] every single word, every syllable, every inflection,” said Travis’ wife Mary, describing the meticulous editing required to get their newest song just right.
The team also needed a “donor voice” — a new singer to provide lyrics that Randy’s AI-generated voice could then sing. They found what they were looking for in James Dupré, a rising country performer who found himself at the center of this unprecedented project.
“I hear Randy — but I also hear me,” Dupré said of the surreal experience of having his own voice layered alongside Randy’s.
Travis’ wife describes it another way.
“It was like a clothesline,” she said of the foundation Dupré provided the song, “and they hung Randy’s voice over it.”
At a recent concert in Pensacola, Travis’ fans were able to hear the new AI song in person. It ended with a standing ovation — and the Travises beaming on stage.
The couple knows they’re at the precipice of an important moment at the intersection of technology and art. Earlier this year, they even testified before Congress on the possible threats of artificial intelligence. Mary warned that this increasingly prevalent tech could lead to the exploitation of artists’ voices — without their consent.
“It feels like identity theft,” she told Congress. “There’s good and bad [in AI]. It just depends on whose hands it’s in.”
If the Travises were going to embrace AI for Randy’s art, they wanted to make sure they checked all the necessary boxes in terms of being comfortable. They wouldn’t have done it without Randy’s full-on approval. And they wouldn’t have been so on board with it if it hadn’t used real vocal stems of his voice.
To say they’re pleased with the result might be an understatement.
“It’s, like, here is this voice that we know — and it’s back,” Lehning said.
And the release of “Where That Came From” is just the beginning. Following the success of its release, Travis and his team are now exploring the possibility of a whole new album along these lines, full of songs that will once again feature the AI model voice — and other donor voices.
“They’re still interested in hearing Randy Travis,” Mary said of her husband’s fans. “So I know I’m not alone.”