Randy Travis’s beautiful baritone was lost. AI helped him sing again.

BOWLING GREEN, Ky. — It’s always devastating when a singer loses their voice, but it felt especially wrong when Randy Travis lost his. The country music icon’s warm, velvety baritone sang songs about love and life that soothed his fans back when he first broke through in the 1980s, and still do. From the first notes of any of his songs - the distinctly gentle twang, the deep tone - you just knew: That’s Randy Travis, unmistakably.

In 2013, Travis suffered a near-fatal stroke that resulted in aphasia, a disorder that took away most of his ability to speak and sing. Yet he was determined to remain in the country music world. After years in physical therapy, Travis started to occasionally reappear at events and concerts around Nashville, where he would delight the audience by singing a few lines of “Amazing Grace,” or deliver the final “amen” of his hit “Forever and Ever, Amen.”

Those cameo appearances may have been enough, but this year, Travis, 65, went back into the studio and back on tour - and in so doing, became the centerpiece of an ongoing conversation about artificial intelligence and artist rights. AI gets so close to replicating Travis’s voice that it has, in a sense, brought him back as a full recording artist. The music industry’s reaction to AI is like that of many other industries: impressive new friend or creativity-destroying foe? In Travis’s case, it provided a glimmer of hope.

Now on his current tour, which played its last 2024 date Nov. 3 here in Bowling Green (and will resume in January in Tulsa), he kicks off the show by walking onstage with his wife, Mary. The two sit in chairs next to the band, almost all of whom are Travis’s original touring musicians. Then James Dupré, a 40-year-old country crooner whose tone sounds remarkably similar to Travis’s, takes his place at the microphone as the surrogate singer, and spends the next two hours singing Travis’s many hits from his four-decade career.

The audience had many reactions: cheers, laughter, as well as sniffling during quieter moments. But the most emotional scene arrived during the performance of “Where That Came From,” a quiet ballad that made headlines when it was released this past spring, because the song was produced using AI, which cloned Travis’s vocals and resulted in his first new song in a decade. According to testimony that Mary Travis delivered on Capitol Hill in June, it’s “the first [AI] song ever recorded and released with full artist consent and involvement in a studio setting,” where his “label, producer, musicians, vocalists and Randy [were] all present.”

During the concert, Dupré sang the first verse and chorus, then stepped away from the microphone - then that famous Randy Travis voice rang out over the speakers for the second verse. The audience roared. AI was more than welcome here.

“I think it was spectacular. It makes me tear up, because I’m like - ah, I might tear up now,” said Rachel Owens, a longtime Travis fan whose eyes grew misty after the show concluded. She made a five-hour drive from Dayton, Ohio, for the concert. “Can you imagine being him, who can no longer use his voice to do what he’s done his entire life - but be able to hear his voice again?”

Country music prioritizes career longevity, and if you treat your fans like friends, they will generally stay loyal to you forever. Travis, a North Carolina native, built up a ton of goodwill with country fans over the years. His first album, 1986′s “Storms of Life,” resulted in No. 1 hits “Diggin’ Up Bones” and “On the Other Hand.” Less than a year later, he had his first bona fide smash with the swoony love song “Forever and Ever, Amen.”

“When Randy hit, he just exploded,” said Don Cusic, a country music historian and professor at Belmont University. “It just went from zero to 100 in no time.”

Travis (who was not made available for interviews for this article, nor was a representative from his label) is often credited as one of the artists who marked a return to “traditional” country music when the genre started to become more pop-heavy, following the juggernaut “Urban Cowboy” craze that swept America after the 1980 movie. This sparked a long relationship with his label, Warner Music Nashville, where he has spent the majority of his career. What he may have lacked in flash, he more than made up for in authenticity. The voice was everything.

“How do you describe a thing that just hits you in the center of your chest and unconsciously makes you feel full and familiar and known?” Cris Lacy, the co-chair and co-president of Warner Music Nashville, said in a May “CBS Sunday Morning” interview, when she was asked what made Travis’s voice was so unique. “How do you describe something like that?”

The idea for the AI-generated song came from the label, Lacy told CBS: “We started with this concept of: what would ‘AI for good’ look like for us? And the first thing that came to mind was: ‘We would give Randy Travis his voice back.’”

AI, of course, has been a contentious subject in the music industry; on Friday, the Beatles’ “Now and Then” became the first song using AI to land a Grammy nomination. Nashville has been on high alert. In March, Tennessee became the first state to sign into law the ELVIS Act, which protects musicians if someone tries to replicate their voice using AI without permission. A few months later, record companies including Sony Music Entertainment, Universal Music Group and Warner Records (Travis’s label’s parent company) sued two AI music generators.

In May, when Warner Nashville announced the release of the AI-produced “Where That Came From,” they launched it with a careful press rollout that emphasized a message: Although they were aware that there is lots of skepticism around AI and its abilities, Travis was fully on board and excited and in the studio for the production process. As described on the “CBS Sunday Morning” story, the song was built from 42 samples of previous Travis vocals gathered by the singer’s longtime producer, Kyle Lehning, which were then fed into an AI program.

That was layered on top of Dupré's vocals for he track, written by John Scott Sherrill and Scotty Emerick; Dupré had recorded the song in 2011 but it was never released. (Dupré was given a credit of providing a “vocal bed” for the song, and the label told Rolling Stone magazine he was paid for his involvement.) After months in the studio, “Where That Came From” sounded just like Travis’s voice.

“Randy’s on the other side of the microphone. It’s still his vocal. There’s no reason he shouldn’t be able to make music,” Lacy said. “And to deprive him of that, if he still wants to do that, that’s unconscionable.”

Ben Camp, an associate professor in the songwriting department at Berklee College of Music and an AI faculty expert, said the proliferation of AI songs is inevitable, and predicted there will eventually be a way to toggle between “human-written music” and “AI-generated music” on streaming services. One important factor is to make sure that people who trained the AI models or wrote songs to train the models are fairly compensated, Camp said, not to mention artist consent.

But when Camp saw the news coverage about Travis’s song, there was “this feeling of excitement and surprise.” While there are still a lot of questions around AI, and some worry that the early iterations could open a Pandora’s box if it’s not regulated, this situation stands out.

“I was touched that an artist who has had such a long and fulfilling career, who just had his ability to communicate with his audience immediately cut off through no fault of his own … that there’s a technology that allows him to connect with those fans,” Camp said. “I’m especially glad that it was done with Randy’s blessing himself rather than someone else making the decision for him.”

Laurie Stras, a professor of musicology at the University of Southampton who also specializes in music and disability studies, said Travis’s shows can both appeal to the fan nostalgia and benefit Travis, who may not have initially been able to envision going back on tour.

“When you have an artist who becomes disabled or an artist who is already disabled and … using technology to create music that they would not otherwise be able to create, we do see this as a positive thing,” Stras said. “Wanting to speak authentically and with the voice that you can call your own is something that is really important to many disabled people who need some kind of vocal assistance.”

Travis still does meet and greets on this tour; backstage, people flocked to him for pictures and to sign CDs. During the concert, he happily engaged with fans as he sat near the edge of the stage. He waved and pointed, and when people surreptitiously left their seats and walked to the front of the stage to take a photo with Travis right behind them, he smiled at the phone cameras. He held hands with his wife as they watched Dupré perform Travis’s string of hits. During “Forever and Ever, Amen,” Travis sang “Amen” into the microphone.

Throughout the songs, a video screen featured clips from Travis’s career, including some of his old music videos, live performances, acting appearances (those “Matlock” episodes!) and even scenes from his old Folgers and Cola-Cola commercials. Mary Travis took the microphone several times to talk about Randy’s legacy and how much they adore Dupré.

“We’re tickled to death that James would come out and do this with us so that he could sing Randy’s great music to you and we would all get to be together again,” she said. “Randy would rather be here than anywhere else in the world, just listening to music and being with his fans.”

The concert also played a video that explained the timeline of Travis’s stroke and the origins of the AI song, and acknowledged the initial skepticism; it then featured comments that showed how fans changed their minds when they knew Travis was part of the process and they got to hear his voice again.

Beth Ziesenis, who runs the website Your Nerdy Best Friend, attended the tour stop when it came to the Ryman Auditorium in Nashville in June, and loved seeing the joy on Travis’s face as he watched the show. As someone who delivers keynote speeches about AI, she was fairly shocked at how accepting the audience was about AI - even with the engineering and production wizards that work in Nashville studios, country music does not have a reputation of being ahead of the curve with technology.

“That was just a major indication of the acceptance of voice cloning and the acceptance of continuing the legacy of some of our iconic singers,” Ziesnis said. “Here we are in the mother church of country music with one of the most traditional country singers with some of the most traditional country fans. … Of course it had Randy Travis’s stamp of approval, but it was just like, ‘Wow, are artists going to embrace this?’”

It remains to be seen how singers will embrace the technology (the CBS segment included country stars such as Carrie Underwood and Cole Swindell growing emotional as they listened to Travis’s new song), though it does raise more existential questions: Will there be a groundswell of estates using AI to bring back the vocals of deceased singers? Is there room for growth of new music if nostalgic listeners only want to hear music from legacy acts? Travis’s family said they embarked on this new music to inspire others. “It’s not what you take when you leave this world behind you,” goes one of his most popular songs, “Three Wooden Crosses.”

“It’s what you leave behind you when you go.”