A few weeks ago, we took a look at how people were using generative AI tools to create deepfake photos and win art competitions. Since then, a rapid acceleration in generative AI music tools has artists and music labels worried. Online tools like Boomi and VoiceMod are giving amateur creators the power to generate music that sounds just like the real deal.
Just like chatbots, it’s causing a stir with some artists who are definitely concerned AI generated music will be used as a substitute for their own music. On today’s deep dive, we’re taking a look at the rise of a new breed of composer.
“There’s a lot of artists that are curious, but I think there are some that are nervous.”
To start off, how do these tools work? I spoke to APRA, who represent 100,000 songwriters, composers and music publishers in Australia.
“In order for generative AI technology to work, it needs a preexisting data set.”
We’re talking about sophisticated algorithms which are trawling the Internet and analysing music tracks. Along the way, it identifies patterns across different genres, melodies, instrumentation, harmonies and structures and learns the intricacies of the music it analyses.
In simple terms, it’s able to reproduce what Kanye or Taylor Swift sound like because it has mastered a deep understanding of the intricate sound of their voice. Things like their characteristics, their inflections, and then it can essentially imitate them. But it also begs the question whose consent has been given in terms of the music that is being used to train these systems.
“A lot of composers live off production music. They spend their time working on, let’s say, an album of trailer music, ambient music… Their music is put out there and if somebody licences their music into a project, they get royalties.”
“Some people really depend on this solely for their income, and I think this is one of the first things that could potentially be replaced by artificial intelligence.”
And while some musicians are concerned, others, like Grimes, for instance, are jumping on the front foot, allowing users to create new music with her voice as long as they give her a cut of the earnings. And just how good is a Grimes endorsed AI song exactly? Turns out not so bad. Speaking to the New York Times, Grimes said that she thought the chorus was really good and could be convinced that she herself had actually worked on it. While these tools are powerful and disruptive, generative AI isn’t the first bit of innovative tech to challenge the music industry.
Back in ‘99, Napster shook up the music industry by creating a platform that allowed people to freely share and download pirated or ripped music online, stifling physical album sales.
That was the first time that the music industry, in particular the record labels, were challenged or had lost control of the distribution. Despite legal action that would eventually shut down Napster, its impact was significant, setting the stage for the digital music revolution and changing how people listen to music forever.
The recent announcements from Spotify suggest that our listening behaviours are set for yet another change. Building off their personalised and curated features like Discover Weekly, Spotify have reinvented the radio DJ in a way with conversational commentary that plays out in between tracks.
“We’re now in an environment where the algorithms are making very personalised recommendations. If you rely on music being pushed to you in that way, it does take some of the exploration and potential for discovering amazing new music. The other concern is that a lot of those playlists and algorithms are generated overseas and the amount of Australian or New Zealand local music that features is very low.”
“I think in the media landscape, writing music that can be synced to picture, that’s a decent source of income. And so some media composers are worried about the tools and how they’re earning potential is being chipped away by tools like this.”
“ If a company can easily generate music material at the touch of a button, then what does that mean for people that have been trained for many years to do this? So of course there’s a bit of angst.”
While the industry has been reluctant to embrace new technologies in the past. No one flinches when they hear a drum machine or Auto-Tune used in music production today, and most consumers are very happy to stream unlimited music from their phones.
“Do you think that there will ever be an AI music category in the APRA Awards one day?”
“For copyright exist in a song, there needs to be a human author element, but it wouldn’t surprise me if a songwriter stands up on stage winning an award and says, I used this tool to help me create this work… I wouldn’t fall off my chair in surprise if that happened.”
“Maybe I have to worry in like another 20 years or something about A.I. being so advanced that it will become the norm of replacing composers.”
Scenes out of Black Mirror, where an AI Miley Cyrus is exploited by their record label might be just around the corner, introducing Lil Miquela, an Internet personality who’s actually completely virtual.
“She has this enormous following on Instagram and there is no doubt that the music industry is watching it carefully.”
“Technology has always disrupted and enhanced the creation and business of music, but it does require careful policy making around it.”
The race between music, AI and human composers is on.
“I think if big companies like Spotify find a way to save money by creating new content rather than being an aggregator, they will for sure.”