The place of A.I. in the creative field sparked wide discussion when German artist Boris Eldagsen declined the award in the Creative Open Category of Sony’s World Photography Awards last month, admitting to using the recently on-trend A.I. generative technology to create the award-winning photo.
As the application and development of generative A.I. show no signs of slowing down, discussions about the technology only keep coming — one of the recent topics has spotlighted A.I. generated music when one song ‘featuring’ rapper Drake and singer The Weeknd became a hit on social media, because it sounded exactly like the singers despite the absence of their involvement.
Titled ‘Heart on my Sleeve’ and initially created by TikTok user Ghostwriter977 using A.I. voice cloning programmes, the deep-fake was first seen on social media and uploaded to multiple platforms like Spotify, Apple Music, and YouTube.
The song was reported to have streamed over 600,000 times on Spotify alone before being removed after a statement from the label of Drake and The Weeknd, Universal Music Group, expressed its concerns over the violation of “the rights of artists and other creators.” However, the song was later re-uploaded to YouTube.
A.I.-generated music of this kind is created using a machine-learning algorithm. With the vocal data of specific artists gathered through text and data mining from sources like music streaming platforms, their voices can be cloned as models with just a few clicks.
Once the voice is ready, the creator has one of two options to create a deep-fake song: either use an existing song but change the vocals (like a song cover), or edit the beat and lyrics — the latter of which can be generated as a completely new song with A.I. using its text-to-voice service as the vocals.
While there may be fear of losses from music labels or artists whose voices are being cloned to create songs, there are those who are embracing the advancement of how A.I. can do the work.
Canadian singer Claire Boucher (aka Grimes, or the mother of Elon Musk’s two children), made and offered her own A.I. programme for the public to create songs using her voice while promising the profit split for successful cases.
Meanwhile, David Guetta, a DJ and 'Producer of the Year' at the 2023 BRIT awards, also supports the use of A.I. in music. In an earlier article with BBC, Guetta said he’s sure “the future of music is in A.I. but as a tool”, as "every new music style comes from a new technology".
He also made and played a rave-style song using the A.I.-cloned sound of rapper Eminem in his DJ set. While the song, as Guetta said, isn’t going to be used commercially, this is proof of how the use of A.I. generative songs can be applied to international stages for entertainment purposes.
Did you know the earliest A.I. involvement in music can be dated back to the 1950s?
The first song created from computer-assisted composition was The Illiac Suite created by Lejaren Hiller and Leonard Isaacson in 1957 using the ‘supercomputer’ ILLIAC I. David Bowie also created his own lyrics-generating software, The Verbasizer, in 1995.
Apart from these, the contemporary use of A.I. in the music industry has been around for purposes such as music producing, sound changing, mixing and mastering, and lyric generating.
With A.I. usage on the rise, big technology companies have also been investing in A.I. for music, some resources are even open to the public.
Spotify debuted its A.I. DJ early February, helping users in selected regions to create playlists according to certain moods. The new feature will deliver a lineup of music alongside commentary; users can express their preferences for tailored adjustments and suggestions.
Meanwhile, Google announced on 10 May its A.I. music generation tool, MusicLM, opens to the public to enable users to create music using text — simply type in some simple descriptions of the genre and the mood to produce a song in a matter of seconds.
In fact, nearly 20 A.I. music models were released in 2023 alone, offering a variety of functions for users to select from.
As the development of A.I. will only continue to grow, it’s expected that the products of A.I. in the music industry will be seen more often (i.e. The last part of the article). With easier access, music production is predicted to be popularized but it also leads to debates on the technology.
With the technology available to produce music with clicks away, even amateur users will be able to produce quality music if used properly. For those who have not yet discovered their talent in music production, this’ll be a great opportunity to give it a go, revealing the artistic side of them.
It’ll also be beneficial creatively to leverage the power of A.I.. The blend of A.I. creations in music, video, and graphic can lead to next-level entertainment filled with countless possibilities.
On the other hand, some may think the popularization of music production will lead to the market being filled with a huge amount of unoriginal and mediocre songs. And as the pool expands, it’s harder for good songs to be discovered.
The lack of human involvement in music production comes as another concern — when music is produced by machines, some say they may be unauthentic and emotionless, neglecting how one of the purposes of music is to express human emotions.
Above are only some examples and the list only goes on and on. But as music listeners, maybe we can sit back and just take this chance to get exposed to what the new technology offers or even try producing the songs of our own with all the options available in the market.
Apart from the recent hit, ‘heart on my sleeve’, there are hundreds, if not thousands, of other A.I.-generated songs available online using voice cloning. Here are some examples songs using A.I.-cloned voices of most renowned artists and even politicians: