Artificial intelligence, or AI, has made it possible to recreate classic songs with different artists like Rihanna and Kanye West performing them.
There is a remarkable rendition of Beyoncé’s “Cuff It” by Rihanna, but the big twist is that Rihanna was not the one singing it. It was an AI program.
In fact, one user used Rihanna’s voice to sing numerous tracks by other artists, including Drake’s “Find Your Love,” Katy Perry’s “I Kissed A Girl,” Maroon 5’s “This Love,” and Omi’s “Cheerleader.”
View this post on Instagram
The Shade Room has collected several samples of Rihanna’s what-if songs, and the sound of the performances is objectively smooth, with only a few static glitches throughout the piece.
It has not just Rihanna’s voice that has been digitally imitated online.
In mid-March, the voice of Ye, the artist formerly known as Kanye West, was the subject of several AI-generated memes.
Using the power of AI tools, Ye can be heard singing original soundtracks from cartoons such as Spongebob and Shrek’s All-Star.
The songs, eerie and jarring to some, are meant to be enjoyed as parodies. Ye’s actual musical style is difficult to pin down, as it has evolved significantly over the course of his career.
Ye performing a song intended for children’s TV and film productions with the same energy he brings to artistic and experimental songs is hilarious.
A popular AI cover created by users features male artists like Ye, Kendrick Lamar, and Drake rapping over Ice Spice’s “Munch,” a song about a woman who is tired of being pursued by men who are only interested in her body; a “munch” is slang for a man who wants to perform oral sex on a woman.
AI covers are created using a variety of techniques, but a common tutorial instructs curious users to download their desired song from an original artist, separate the vocals and beats using software called X-minus, and follow a Google Docs machine learning research community.
Machine learning is a type of AI that allows computers to learn without being explicitly programmed. In the case of AI covers, machine learning is used to train AI models on large datasets of music. These models can then be used to generate new music that sounds like the artists used to train them.