Randy Travis, who lost much of his speech in a 2013 stroke, used artificial intelligence technology to clone his voice for his first recording in more than a decade.
Travis, his longtime producer Kyle Lehning, Travis’s wife Mary, and Warner Music Nashville co-chair and co-president Cris Lacy spoke with CBS Sunday Morning to detail how AI helped create “Where That Came From,” Travis’s new song that released on Friday. The full report will air Sunday.
Lehning confirmed that another singer performed the initial vocals before overlaying the raw performance with Travis’s voice clone. As of this story’s publication, it’s unclear who the other singer is.
“It’s not about how it sounds. It’s about how it feels,“ Lehning told CBS. “Him being here and him being able to be, you know, a vital part of the decision-making process makes all the difference to me.”
“Where That Came From” was co-written by country veterans John Scott Sherrill and Scotty Emerick, per a citation on a YouTube video previewing the song. Emerick released the song himself last year on his EP Headwinds (The Demo Sessions).
Travis suffered a near-fatal stroke in 2013. Since then he’s released two albums of previously recorded country covers and has made sporadic appearances onstage with other artists to sing the closing “amen” of his hit “Forever and Ever, Amen.” His new song is one of the first commercial releases to feature AI-cloned vocals, a landmark moment for AI music, which has become one of the most pressing topics in the music industry. In the past year, artists, record labels, and lawmakers have grappled with how to regulate the tech to ensure protections for artists’ works and their digital likenesses.
“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.”
Much of the industry has said that AI could be immensely helpful in the creative process if the proper guardrails are put in place. For example, earlier this week fellow Warner artist FKA Twigs revealed in a testimony to a Senate judiciary panel that she developed an AI deepfake of herself that could interact with fans on social media and let her focus on making music.
The tech has proven controversial when the artists themselves aren’t involved and their recordings are training AI models without their permission. An anonymous songwriter going by “Ghostwriter” set the industry aflame last year when he released the song “Heart on My Sleeve” featuring AI-generated vocals of Drake and The Weeknd. The artists’ record company Universal Music Group pushed for streaming services to remove the song.
A year later, Drake released his Kendrick Lamar diss track “Taylor Made Freestyle” featuring the AI vocals of Tupac. The late rapper’s estate issued Drake a cease and desist, calling his song a “flagrant violation of Tupac’s publicity and the estate’s legal rights.” Drake subsequently took the song down.
Travis’s AI song brings forward several questions, such as how it’s handled for copyright and credits. It isn’t immediately known how the singer who recorded the vocals is cited or compensated, or what type of license, if any, was needed to get Travis’s AI vocals over the line.
And as many remain skeptical or even afraid of how AI should be used, the new song may be a significant conversation starter.
“Randy’s on the other side of the microphone … It’s still his vocal,” Lacy said. “There’s no reason he shouldn’t be able to make music … And to deprive him of that, if he still wants to do that, that’s unconscionable to me.”