Last post, I talked about how VALL-E and voice synthesis, the artificial cloning of one’s voice to create synthetic speech, is profoundly impressive. One industry where such technology has already proven to be controversial and greatly impactful is music.
Many famous artists have great power through their unique and recognizable voices. Voice synthesis allows anyone to use artists’ voices for their own purposes. Anyone can decide to create a song featuring a popular artist without disclosing that it’s not real. They see a lot of benefit from this as the song can gain much more traction due to the artist’s name being attached to it. This is a harmful form of plagiarism because artists that don’t consent to having their voices used don’t gain anything from it despite the role their fame plays in the song’s success. At the same time, it is a very nuanced issue as it is unclear whether or not an artist or label “owns” the rights to the synthesized voice and deserves to be paid royalties in a similar fashion as sampling music does. The question of how much royalties an artist deserves for the use of their voice through artificial intelligence is even more complex. While having a popular voice on a song may contribute to more success, it is arguable that it’s ultimately the talent of the songwriter or producer that allows a song to become popular. Advanced voice synthesis technology is a gray area for music, but it’s already making tremendous ripples.
Sincerely, Aubrey is an album featuring names such as Drake, Lil Baby, Kendrick Lamar, Kanye West, Travis Scott, Rihanna, Future, and The Weeknd. Every track is created by mixing existing vocals of every artist with Drake’s, which is entirely synthesized. The album is shockingly good and would’ve easily fooled me had it not been made clear that AI was used. However, the response from Universal Music Group (UMG), the music company that represents Drake, was to immediately urge streaming platforms to block all songs featuring AI versions of their artists, claiming that it violated the rights of their artists. Regardless of whether or not this constitutes copyright infringement, it’s extremely difficult to stop the practice entirely due to the nature of music as developers have access to a lot of artist’s voices to be used to train their voice synthesis models. This poses additional questions as to how to prevent AI music, if at all, and how to ensure that listeners are aware of the source of the content they consume.
While the potential of harmful plagiarism through the use of voice synthesis is real, there are also some benefits that AI provides to the music industry. For one, artists and producers can test out different sounds and lyrics without having to record expensive studio sessions. This saves money while increasing production speeds. In addition, a common practice in music is for a songwriter to pitch a demo of a song to a record label to consider or to an artist to sing it themselves. With voice synthesis, the label or artist can actually listen to what the song might sound like without wasting the artist’s time.
It is undeniable that voice synthesis can facilitate plagiarism in the music industry through the creation of AI music. However, it is also notable that AI music can be largely beneficial to artists and producers.
See more like this:
