Randy Travis returned to recording on May 3, 2024, with “Where That Came From,” his first new recording in more than a decade. The country ballad uses AI-assisted vocal production to recreate the sound of Travis’s singing voice after the 2013 stroke that left him with aphasia and severely limited speech. It was an authorized, human-directed project—not a song written and performed autonomously by AI.
What was Randy Travis’s return?
Warner Music Nashville released “Where That Came From” on May 3, 2024. The label described it as Travis’s first new recording in more than a decade. The date matters: this was a 2024 release, not a new 2026 development. Warner Music Nashville’s announcement introduced the song and explained the circumstances behind it.
| # | Preview | Product | Price | |
|---|---|---|---|---|
| 1 |
|
I Told You So - The Ultimate H | $10.91 | Buy on Amazon |
| 2 |
|
Randy Travis: The Very Best Of | $18.25 | Buy on Amazon |
| 3 |
|
Greatest #1 Hits | $4.00 | Buy on Amazon |
| 4 |
|
Worship & Faith | $17.06 | Buy on Amazon |
| 5 |
|
Hymns: 17 Timeless Songs of Fa | $10.99 | Buy on Amazon |
Travis is a Grammy-winning country singer and Country Music Hall of Fame inductee, known in part for a distinctive baritone that helped define his public identity. After his 2013 stroke, he developed aphasia and his ability to speak and sing was severely limited. The new recording made it possible to hear a recognizable version of his singing voice again, without implying that his physical ability to sing had returned.
How did AI help make the recording?
Travis’s longtime producer, Kyle Lehning, worked with archived vocal material to create a version of Travis’s voice for the song. The Associated Press reported that the team built two AI models: one using 12 vocal stems and another using 42 stems, drawn from recordings spanning approximately 1985 to 2013. These figures describe the reported source material for this project, not a standard recipe for voice reconstruction. The AP account describes the models and collaborators involved.
#1 Best Overall
- I Told You So: The Ultimate Hits of Randy Travis
- Brand: WARNER RECORDS
In plain terms, the system learned from examples of Travis’s recorded voice and helped render vocals that resembled his singing. That is different from placing an untouched, newly recorded Travis vocal on the track—and different from asking AI to create an entire song by itself. Songwriting, arrangement, production choices and oversight remained human work.
What AI did—and did not do
- It did: use authorized recordings as examples of Travis’s voice and help produce a vocal performance resembling his recognizable sound.
- It did not: restore Travis’s health or physical singing ability, nor is the song reported as an autonomous AI composition or performance.
The song was written by Scotty Emerick and John Scott, according to the Los Angeles Times. The release was accompanied by a lyric video and behind-the-scenes coverage.
Rank #2
- Very Best of Randy Travis by Randy Travis (2004-08-03)
Did Travis approve and participate in the project?
Yes. Warner approached Travis and his wife, Mary, about the idea; the official announcement and reporting by the AP and CBS describe their involvement and support. Travis was present during the project and expressed enthusiasm about making music again. In the label’s announcement, he described the recording process as special and said he was excited to share the song.
That participation distinguishes this release from an unauthorized imitation that uses an artist’s voice without permission. The team’s stated purpose was to help Travis continue creating, using recordings of his own voice, with public disclosure of AI’s role. Those facts do not settle every question about voice models, but they matter when assessing this particular project. CBS News’s behind-the-scenes report documented the process and its personal significance.
Rank #3
- Shrink-wrapped
Why did the song feel emotional to collaborators and listeners?
The emotional force came from more than hearing a familiar country voice. Travis’s voice had been central to his career, and the stroke had sharply changed his ability to perform vocally. The recording offered him and the people around him a way to hear a recognizable version of that voice in a new musical setting.
CBS’s coverage showed emotional responses to the song and to the process of making it. Later, the recording was discussed in a congressional hearing about music and copyright, including its emotional impact. Neither response means every listener welcomed the result: the release also prompted questions about authenticity and AI’s place in music. The hearing transcript records the later discussion.
Rank #4
- 901001643
- 9780901001641
- 080688627324
What does the project show—and what remains unsettled?
“Where That Came From” shows one way AI voice technology can assist an artist who knowingly authorizes its use: it can help collaborators make a new recording when the artist’s voice is no longer available for an ordinary performance. It does not establish that the same process will work for every singer, that artists generally accept AI-generated vocals, or that this project creates a legal precedent.
The case also leaves broader questions open. A simulated voice may sound familiar without reproducing the choices, interpretation or emotional nuance of a live performance. Listeners may need clear disclosure to understand what they are hearing. And one artist’s consent cannot stand in for consent from other performers, estates or rights holders. A successful, authorized project therefore should not be treated as permission for labels or others to clone voices without equivalent agreement.
Best Value
For more on the wider debate over authenticity and voice replication, see The Washington Post’s later analysis.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
