Randy Travis Sings Again - with a little help from AI
With all of the negative attention that AI in music has been getting lately, here’s something that should brighten your day. Country music legend Randy Travis, known for his distinctive baritone voice and numerous hit songs, has been largely absent from the music scene since suffering a stroke back in 2013. The stroke left him with limited vocal abilities, making it all but impossible for him to record new music. That all changed recently when his long-time producer Kyle Lehning and Cris Lacy from Warner Music Nashville approached Randy and his wife Mary to try to use AI technology to recreate his voice on a new song.
The AI-assisted song, titled Where That Came From, was released about a month ago, and it was created using a combination of machine learning algorithms, specifically an amazing AI voice modeling tool called My Vox, and Travis's extensive catalog of previous recordings. By analyzing Travis's vocal patterns, phrasing, and stylistic nuances from his earlier work, My Vox was able to generate new vocal tracks that closely mimic his signature sound. They also brought in another country music star, James Dupree, to actually record a demo of the vocals which then served as the track that the Ai-generated version of Randy’s voice was laid over. The end result is pretty incredible. Here’s the song, Where That Came From.
What I LOVE about this project is that in this unique situation, an AI tool was utilized to analyze and then recreate the voice of an artist who no longer has access to that instrument. I watched a segment on my all-time favorite TV Show, CBS Sunday Morning, a month or so ago and had a combination of a huge smile and a lump in my throat throughout the segment. To see technology used as a force for good, and to see the reaction of not only Randy himself, but of the other country musicians, recording engineers, producers and record company executives stunned at the end result and beaming ear to ear, was very special. Here is that segment from CBS Sunday Morning:
This story comes right on the heels of the recent announcement of a partnership between Universal Music Group (UMG) and SoundLabs to allow musicians to create AI-created clones of their own voices for their own use - unlike sites like Jammable that allow you (or used to allow you) to use artists voices for your own use. In my opinion, technology companies like SoundLabs and My Vox should run towards these types of partnerships to avoid lawsuits. If there is a way for both the artists and the tech companies to co-exist peacefully and make money at the same time, I’m certainly all for it.
In the meantime, if you want to take the My Vox AI technology for a test drive, you can actually create a model of your own voice for free just by signing up for a free account. This type of technology would be very cool to use with students who might by shy about singing in front of anyone - or even for yourself. If you do try it out, let me know what you think!
