Universal Music Group partners with AI voice-cloning company

Rolling Stone reports that Universal Music Group has just inked a deal with a new AI music tech startup called SoundLabs. From the article:

UMG's artists and record producers will be able to use SoundLabs' upcoming feature called MicDrop starting later this summer, and as the companies said in their announcement, the platform allows the artists to make voice models of their own using data the artists provide. SoundLabs gives the artists control over the ownership and use of the voice models, the companies said, and the voice clones won't be made accessible to the general public.

Aside from merely making a copy of a voice, MicDrop purports to offer a voice-to-instrument function, similar to the features that can make keyboards sound like a guitar or drum. MicDrop also offers language transposition, the company said, which could help artists release songs around the world without a language barrier.

Earlier this year, UMG got into a pissing match with TikTok. Part of the issue there involved AI, as Pitchfork reported at the time:

Regarding AI, UMG asserts that, despite its attempts to protect the earnings of songwriters, TikTok is "allowing the platform to be flooded with AI-generated recordings—as well as developing tools to enable, promote and encourage AI music creation on the platform itself—and then demanding a contractual right which would allow this content to massively dilute the royalty pool for human artists, in a move that is nothing short of sponsoring artist replacement by AI."

So it seems here that the issue is not about AI-generated voices creating artificial music — but rather, which multinational corporation gets to profit off of it. In that particular situation, UMG was not satisfied with TikTok's assertion that the record label should essentially allow voice-cloning "for exposure," as it were. But if UMG can circumvent the labor costs of compensating artists by using AI voice cloning technology to produce new music "content?" That's a big hell yeah!

As Ryan Broderick of Garbage Day accurately observed:

It's a bit of a chilling thought, but these next few months, before this plugin rolls out, are basically going to be the last time we can kinda-sorta be sure that what we're hearing is from a human being and not AI-generated.

[…]

The "autotune" sound that we all think of is actually what happens when a musician dials up the settings all the way and it starts to degrade the audio. So it's really hard to predict how these AI models are going to start influencing music once they make their way into producers' hands.

And there's an interesting point there. Autotune technology is not inherently bad; it can actually be quite helpful, in a lot of different situations! But it got a bad reputation* because some human artists —particularly rappers, perhaps T-Pain most infamously — leaned into the artificial nature of the technology. They embraced the digital distortion as an effect, using it turn their voices into a different kind of instrument. Since then, the distinct sound of that pitch-shifting effect has become much more normalized, for better or for worse.

But at the end of the day, that autotune sound still represents a human voice — which is to say, it's not so different from plucking a guitar string, converting the vibrations into an electromagnetic frequency that gets transported down a wire and then filtered through a similar phasing or fuzzing or pitch-shifting effect. There's something analog at the source of it. And soon enough, we might be longing for that authenticity — or maybe we won't even be able to tell the difference. Maybe human musicians will start leaning into and embracing certain kinds of analog distortion in order to send a signal that this is authentic human music. Of course, I'm also saying this as someone who sort of wants to embrace more digital guitar amp/effect technology, but has spent so much time and money investing in analog gear that it feels like it's too late.

Either way: what a weird time to be alive.

The World's Largest Music Company Is Helping Musicians Make Their Own AI Voice Clones [Ethan Millman / Rolling Stone]

*Whether that backlash was about discomfort with the uncanny valley-ness of it all, or if it was racially motivated, or a broader curmudgeonry around "kids these days and their loud music," that's a more complicated topic for another time.