June 6

Does Randy Travis’ AI voice make cloning ok?(Voice Cloning Part Two)

Editor’s Note: this post IS about technology, but I have to give some back story first. Hang in there.

This week my husband and I scored tickets to see Randy Travis at the mother church of country music, the Ryman Auditorium. (If you don’t know who country music legend Randy Travis is, stop reading right now and watch this.)

A stroke in 2013 robbed Randy of his velvety voice, so this concert was different. Randy and his wife, Mary, chose a former Voice contestant named James Dupré to sing Randy’s tunes on stage while Randy and his wife sat on the side in armchairs.

James’ voice is eerily similar to Randy’s, and he sang the songs exactly the way I remembered them. (All this is relevant to the tech part… hang in there.)

The night focused on Randy’s top 16 hits. Wife Mary acted as narrator/emcee throughout, offering commentary and introducing guests. Behind James, a big screen showed clips from Randy’s videos and performances. Most of Randy’s original band backed him up, and I was over the moon with nostalgic, twangy, country happiness.

Sometime after “Digging Up Bones” and before “Forever and Ever, Amen,” the video screen started what would be a 5-minute mini-documentary about the making of Randy’s just-released song, “Where That Came From.” And that’s when I stopped singing and started recording.

Randy’s AI Voice

The video told the story of how a team of traditional music professionals and newfangled AI experts worked together to clone Randy’s voice. Then they had James sing a new original and overlaid Randy’s sound. The results were astounding.

It was truly — almost — kind of — nearly — Randy Travis.

How Randy Travis cloned his voice

A couple of weeks ago I did part one of a voice cloning series (stay tuned for at least one more). I showed you how I could use one voice recording and apply another one over it with creepy success. I did this with software that costs about five bucks a month, but it sounds like Randy’s process was similar. Here’s a clip from CBS Sunday Morning about the whole process.

Is this a justifiable use case for AI voice cloning?

After the other post, people are left wondering if AI clones will literally steal people’s voices, scam everybody and generally destroy the fabric of modern society. Beats me, but it seems like Randy Travis’ voice is a success story. His wife cried several times during the concert, saying how much she loved his songs and missed hearing him sing them. But now he can share his music again as much as he wants. That sounds like a win.

But on the other hand (Randy fans… see what I did there?), isn’t there something special about being a part of the natural cycle of life for an artist? If every artist can keep producing songs in their voice long after they no longer sing the notes themselves, is it still art? What happens when the charts are filled with songs from artists long gone? Is there a place for new artists when the old ones don’t go away?

Please Watch the Fans’ Reactions

When I realized that they were talking about the voice cloning, I started recording. This is officially the very first bootleg video I’ve ever published. My apologies for the bad camera work, shaking, awful audio and perhaps the violation of Randy’s intellectual property and copyright (note: It did get flagged as a copyright violation from YouTube but they said the copyright owner allowed its use).

I really want you to watch it to be in the moment with the fans… to see Randy’s face… to hear his wife talk about what it means to them. This snapshot of an AI voice’s public release tells me a lot about the motivation behind some artists and the fans that support them.


artificial intelligence, audio, ethics, security

You may also like

Make your own graphic alphabet

Make your own graphic alphabet
{"email":"Email address invalid","url":"Website address invalid","required":"Required field missing"}