Impersonation is the Highest Form of Flattery, or is it? The Danger of AI Generated Music
From AI generated essays to art, this technological revolution has and will continue to create many legal issues for its engineers and affected creatives alike. This article will focus on issues presented by AI generated music, which is music created using algorithms and computer programs which create new pieces fully on their own.1 Some recent examples in today’s media include DJ David Guetta using Eminem’s voice to create a completely new song that sounds like it was performed by Eminem himself, and SZA’s “Kill Bill” being altered by AI to sound as though it was Ariana Grande singing a cover. AI generated music has immense power to change the music industry as we know it, but without proper legal protections, artists are at risk of major losses when it comes to the right to their own voice.
The Copyright Protection of Sound-alikes
A sound-alike is a cover of a sound recording intended to be so similar to the original, that the average listener would have a hard time telling the two recordings apart. Within the music business, there exists “replay” companies which make profit off licensing sound-alike covers for uses ranging from sampling to synchronization. These companies describe their services as creating, “authentic sounding replacement productions to help you avoid very expensive, slow and sometimes impossible to get sample clearance” and they often help with necessary publishing clearances.2 While imitating the sound of original recordings may seem like infringement, this is actually a right vested by copyright law.
The U.S. copyright code provides a statutory compulsory license which grants permission to re-record a song that has already been commercially released, and provides rules which allow these covers to be released publicly.3 All that is required is delivering the relevant parties a notice of intent to record the musical composition. The owner of a sound recording copyright has the exclusive right to reproduce copies of their work and create derivatives from it, but this right does not extend to the creation of other sound recordings derived from the same underlying written composition. This allows for such duplication to consist entirely of an independent fixation of other sounds, even though such sounds may imitate or simulate those in another recording of the same song.4 The reason this gap in the rights of those with copyrights in sound recordings exists is that a sound cannot be copyrighted. Although lyrics to a song are copyrightable, the underlying voice is not because sounds are not able to be fixed in a work of authorship.5 As the “sounds are not fixed,” there is no copyright protection available to the infinite number of words or phrases a person might utter in their distinctive voice.6 As the law now stands, cases with a focus on sound-alikes are minimal because it is very rare that a human voice has the ability to sound so much like someone else’s that it would be grounds for a lawsuit. Further, of such cases that occur, most typically settle in avoidance of expensive litigation. One of these rare, litigated cases is Midler v. Ford Motor Co., which lays some important groundwork for the protection of artists in the case of sound-alikes.
Midler v. Ford Motor Co.
Midler began with an ad agency seeking to make a commercial for Ford Motor with the song “Do You Want To Dance” from the Grammy winning Bette Midler. The conversation between the agency and Midler’s manager went as follows, “Hello, I am Craig Hazen from [agency]. I am calling you to find out if Bette Midler would be interested in doing …? manager: “Is it a commercial?” “Yes.” “We are not interested.”7
Midler was notorious for being very strict with her synchronization rights at the time of this case. The ad agency did not take this rejection as a sign that they should seek out other music. Instead of going back to the drawing board, they sought out one of Midler’s backup singers and tasked her with sounding “as much as possible like the Bette Midler record” on a new sound recording.8 After the commercial was aired using the sound-alike created by the backup singer, Midler was told by “a number of people” that it “sounded exactly” like her, and a personal manager in the entertainment business not associated with Midler, declared by affidavit that he heard the commercial on more than one occasion and thought Midler was doing the singing.9
This case is purely about the protection of Midler’s voice as a part of her identity. The ability to control the usage of one’s identity in any context is protected under the right to publicity, which is governed by state law. In the context of Midler, no other aspect of Midler’s name or likeness was used in the commercial and the ad agency had obtained a license for the underlying musical work. As we know, mere imitation of a recorded performance would not constitute a copyright infringement even where one performer deliberately sets out to imitate another’s performance as exactly as possible.10 Under California law, name, voice, signature, photograph and likeness are all categories of identity which are protected. The Court commented that it was made clear that Midler’s voice was of value to the ad agency through them asking Midler to use her voice and after her rejection, studiously acquiring the services of a sound-alike. The Court further stated, “A voice is as distinctive and personal as a face. The human voice is one of the most palpable ways identity is manifested … the singer manifests herself in the song. To impersonate her voice is to pirate her identity.”11 In conclusion, the court held for Midler, finding that when a distinctive voice of a professional singer is widely known and is deliberately imitated in order to sell a product, the sellers have appropriated what is not theirs and have committed a tort in California.12
How to Protect Artists in the Wake of the AI Revolution
When looking specifically at sound-alikes and their intersection with AI, AI is so dangerous because it has made it possible to create phonographs which sound truly identical to an artist in a way that a natural human voice would not be able to. In Midler, the cover song sounded so close to the original because the singer they hired had worked with Midler for years. Now, AI can study any artist’s voice and replicate it exactly within minutes. This has the potential to saturate the licensing market with sound-alikes, which may be licensed for much lower rates than original recordings.
While Midler protected the artist’s voice in the context of sound-alikes, the court was more concerned with the right to publicity/name, image, likeness (NIL) laws in California. Because there is not a uniform federal law defining NIL rights, some states (like California) include voice as a part of identity under NIL and some do not.13 In states that do not have voice as a protected category under NIL, Midler may not have prevailed on the same suit. In order to protect the right that an artist has to their own voice, which is a defining category of identity for vocalists, there must be legislation created that defines voice as a category of NIL on the federal level. This will provide the foundation necessary to give artists the right to fight for their voice in a technological era which threatens it.
Endnotes
- Georgia Carter, AI Generated Music – The Future Of The Fall Of Creativity? (Apr. 5, 2023), https://www.musicgateway.com/blog/music-production/ai-generated-music-the-future-of-the-fall-of-creativity#:~:text=13.3.2023,music%20and%20create%20new%20pieces.
- About Us (Apr. 5 2023), https://www.scorccio.com/replays-audio/
- 17 U.S.C.A. § 115 (West).
- 17 U.S.C.A. § 114(B) (West).
- Butler v. Target Corp., 323 F. Supp. 2d 1052, 1056 (C.D. Cal. 2004).
- Id.
- Midler v. Ford Motor Co., 849 F.2d 460, 461 (9th Cir. 1988)
- Id.
- Midler v. Ford Motor Co. at 461-62.
- 17 U.S.C.A. § 114(b).
- Midler v. Ford Motor Co., 849 F.2d 460, 463 (9th Cir. 1988).
- Id.
- Smith v. NBC Universal, 524 F. Supp. 2d 315, 325 (S.D.N.Y. 2007).