Over a decade after suffering a debilitating and near-fatal stroke, country music star Randy Travis has released a new song that features an AI-generated recreation of his own voice.

It's a fascinating use of the technology. Instead of cloning another artist's voice without their consent or an estate resurrecting a passed performer's vocals, Travis and his wife Mary used the tech to breathe new life into the singer's extensive and decades-long country music career.

The track, called "Where That Came From," features Travis' iconic voice, which was generated by two separate AI models. These models were trained on vocal stems plucked from the singer's lengthy discography, dating back to 1985.

The results speak for themselves: it's practically impossible to tell if it was generated by an AI.

"All I ever wanted since the day of a stroke was to hear that voice again," Mary Travis told The Associated Press.


Travis suffered a stroke, the result of a virus that attacked his heart, in July 2013. He had to spend years relearning how to walk, write, and read. The resulting aphasia still limits his ability to speak to this day.

Warner Music Nashville co-president Cris Lacy came up with the idea to have his AI clone carry on his legacy. She worked with London, UK-based developers to train AI models based on vocal samples.

"I really wish somebody had been here with a camera because I was the first person to hear it," Travis’ longtime producer Kyle Lehning told the AP. "And it was stunning, to me, how good it was sort of right off the bat. It’s hard to put an equation around it, but it was probably 70, 75 percent what you hear now."

The use of AI in the music industry has proven extremely controversial, with major music labels playing a game of whac-a-mole with viral songs that make use of AI-cloned vocals of popular musicians, including Aubrey "Drake" Graham, Abel Makkonen "the Weeknd" Tesfaye, and Kendrick Lamar.

Just last month, Drake was threatened with legal action from Tupac's estate after using generative AI to clone the late rapper's voice in a track called "Taylor Made Freestyle." The Canadian musician ultimately took down the track.

In the case of Travis, however, things look considerably different given his full support for the project.

"There’s just so much chatter about all the negative sides of AI," Lacy told CBS."We started with this concept of ‘What would AI... look like for us?’ And the first thing that came to mind... was we would give Randy Travis his voice back."

Nonetheless, even with full permission, the use of generative AI will likely continue to prove controversial in the industry, especially when it comes to hard-to-answer questions about copyright.

In early April, over 200 artists, including industry heavyweights like Bille Eilish, Nicki Minaj, and Katy Perry, signed an open letter calling on tech companies to stop using AI "to infringe upon and devalue the rights of human artists."

Meanwhile, Travis and his wife hinted at the possibility of releasing more tracks featuring AI-generated vocals.

For her part, Lacy is actively defending the project's authenticity in light of an onslaught of criticism when it comes to AI in music.

"This isn’t a stunt, and it’s not a parlor trick," Lacy told the AP. "It was important to have a song worthy of him."

More on AI music: Internet Fooled by AI-Generated Kendrick Lamar Diss Track

Share This Article