Grimes: AI Can Replicate My Voice With out Penalties

Enlarge / Grimes attends The 2021 Met Gala Celebrating In America: A Lexicon Of Trend on the Metropolitan Museum of Artwork on September 13, 2021, in New York Metropolis.

Theo Wargo/Getty Photographs

On Sunday evening, Canadian musician Grimes tweeted that she would break up 50 % royalties on “any profitable AI generated music” that makes use of her voice, experiences The Verge. As an unbiased artist, Grimes says anybody can use her voice with out penalty. “I’ve no label and no authorized bindings,” she says.

Her announcement comes after controversy over an unauthorized music that includes AI-generated vocals of Drake and The Weeknd that went viral earlier this month. Shortly after its reveal, the music obtained pulled from YouTube and different social media platforms beneath unclear circumstances, which The Verge experiences may have been a botched fan tribute or a advertising stunt.

Grimes is not any stranger to the AI and tech cultural scene, having written music about AI and tweeted concerning the topic steadily. In February, OpenAI CEO Sam Altman tweeted an unlabeled picture of himself with Grimes and AI critic Eliezer Yudkowsky. Grimes additionally shares two kids together with her former companion Elon Musk, who not too long ago signed a letter urging the slowdown of AI mannequin improvement.


Not like Musk, Grimes seems to be all-in on what she sees as optimistic disruption from generative AI. “I feel it’s cool to be fused w a machine and I like the thought of open sourcing all artwork and killing copyright,” she tweeted as a reply to her announcement on utilizing her voice. Later within the night, she wrote that she is engaged on “a program that ought to simulate my voice properly.”

The power to realistically synthesize a human voice utilizing deep studying fashions has grow to be more and more simple over the previous 12 months as a result of open supply fashions similar to tortoise-tts and industrial companies similar to ElevenLabs speech synthesis. We’ve beforehand reported on applied sciences that may clone an individual’s voice from a three-second audio pattern, voice transformation software program, and Disney’s use of AI know-how to synthesize Darth Vader’s voice.

On Monday, Grimes clarified in a sequence of tweets that she is working with her team on a system that may accumulate and pay out royalties utilizing “good contracts,” then she tweeted about doubtlessly issuing copyright takedowns for AI-generated songs with “poisonous lyrics” utilizing her voice. “[I] don’t wanna be answerable for a Nazi anthem except it’s by some means in jest a la producers I suppose,” she wrote.

Nonetheless, she’s conscious of the potential in poor health results of inviting others to make use of her voice with few restrictions and is forging ahead anyway:

“We anticipate a specific amount of chaos. grimes is an artwork challenge, not a music challenge. The last word objective has at all times been to push boundaries quite than have a pleasant music. The purpose is to poke holes within the simulation and see what occurs even when it’s a foul consequence for us.”