Much of the discourse around deepfaked tracks using the cloned voices of famous artists has put musicians in a passive role: this is something that’s done to them, without permission, that they’re powerless to stop.
That’s true, but it’s not the whole story. Some artists have already taken a more active role in trying to shape this technology in a more positive way.
We’ve written about Holly Herndon’s work on Holly+, her own AI voice-clone that other musicians can use in their work – but with a licensing and royalty-sharing structure around it. More recently, Grimes went down a similar path.
Both artists partnered with startups to build their ‘voiceprints’ – Never Before Heard Sounds and CreateSafe respectively – but the musicians were very much the creative driving force.
Now there’s a third example. DJ Fresh of ‘Hot Right Now’ fame – an earworm we make no apologies for planting: it’s a stone-cold banger – has teamed up with music producer Nico Pellerin to launch a service called VoiceSwap.
It’s described as a tool for producers, artists and songwriters who don’t want to use their own voices on songs, enabling them to “use AI to transform their voice to sound like one of our featured artists”.
Liam Bailey, Jamie McCool, Dominique Young Unique, Nikki Ambers and bona-fide house legend Angie Brown are the first artists on board, but VoiceSwap wants more – and says it only needs 25 minutes of their singing or eight acapella tracks to create their clones. Importantly, there are also firm rules around how people can use the service.
“You can use voice-swap to make demos, but you cannot share any voice-swap generated audio publicly, or monetize it in any way without permission from an artist representative who we will connect you to,” notes its website.
“We use state of the art technology to make the AI models output traceable, and the audio is the legal property of the singers. It can’t be used publicly without their permission, just like a real demo made in the studio.”
VoiceSwap is using a subscription business model, with three tiers ($5.99, $9.99 and $29.99 a month) offering progressively more minutes of ‘audio credits’ and other features. People can also try it out by buying 15 minutes of these credits for $7.99. The company says its artists will get a pro-rata share of the revenue it generates from their voice models.
It’s too early to judge whether VoiceSwap will work as a business – musicians will be putting it through its paces to gauge the quality of its technology – and the competition in this sector is sure to intensify in the coming months given how *coughs* hot it is right now.
But that’s a positive thing: not just that there are startups springing up to make licensed, artist-friendly businesses out of AI voice-cloning, but that a number of those startups are being founded and shaped by artists too.