On January 29, AR Rahman posted on X (formerly Twitter) that the families of late singers Bamba Bakya and Shahul Hameed gave permission to use their "voice algorithms" in the music for Aishwarya Rajinikanth's film Lal Salaam. Of course, Rahman isn't the first musician to use artificial intelligence (AI) to recreate something lost to time.
In November 2023, Paul McCartney and Ringo Starr, the two remaining members of The Beatles, used AI to extract John Lennon’s voice from a never-heard-before 1979 demo tape, and George Harrison’s guitar tracks from a failed effort to finish that very song in 1995. In a sense, The Beatles were together again, to produce “Now And Then”—the band’s first song in decades, and their last forever.
Listening to “Now And Then” is an eerie experience. While written as a love song in the 1970s, its lyrics now take on whole new levels of meaning. In its making, it’s also a strange meeting place for the past and future.
“There it was, John’s voice, crystal clear,” said McCartney in a statement that also detailed how they used filmmaker Peter Jackson’s technology which was used to resurrect archival material for documentary extraordinaire Get Back, to achieve this. “It’s quite emotional. In 2023, to still be working on Beatles music…I think it’s an exciting thing,” said McCartney.
“Now And Then” is perhaps a benign example of how 2023 became the year that AI stormed the music world—and immediately polarised it.
Advancements in machine learning have now allowed platforms like Spotify and Apple Music to more accurately suggest music to their subscribers based on their taste.
Elsewhere, it is being used to enhance the process of mixing and production, making it cheaper and quicker. Abundant tools now proliferate that can create loops of beats to create inspiration, that can produce a song based on a text prompt, just like Midjourney could do art. AI can even correct vocal pitch better than auto-tune ever could, but all of this is just the tip of the iceberg.
AI in music hits a snag when it gets mixed in with the actual creation process. Generative AI is built on the sum of all human experience —or whatever of it we’ve been able to upload to the Internet. When you ask a tool that has distilled all of human life into algorithms to create something new, it is bound to imitate more than innovate—at least at this stage. And that’s where the trouble begins.
AI tools for music: downside
In early 2023, Latin pop superstar Bad Bunny declared that he had no intention to collaborate with Justin Bieber any time soon. And yet, in November, a song featuring what sounded like them began to circulate on TikTok, garnering millions of likes. “An artist named FlowGPT had used AI technology to recreate the voices of Bad Bunny, Bieber and Daddy Yankee in a reggaeton anthem... Bad Bunny himself hated it, calling it a 'shit of a song' in Spanish and discouraging his fans from listening, and the clip was removed from TikTok. But many fans of all three megastars loved it all the same,” reported TIME magazine. You can still listen to “FlowGPT Bad Bunny, Justin Bieber Daddy Yankee NostalgIA” on YouTube.
A handle called Ghostwriter went viral for creating a track that sounded exactly like hip-hop gods Drake and The Weeknd, reported TIME; and another creator jokingly set 20th-century American superstar Frank Sinatra’s voice to profane lyrics by crunk progenitor Lil Jon. In this first flush of AI in the music world, mimicry has emerged as a core ethical issue, at least from the musicians’ point of view because it highlights the tricky terrain of ownership, authenticity and duplication.
“I believe that AI should not mimic a person,” said hip-hop producer and Black Eyed Peas founder will.i.am on a podcast in October. Will.i.am, also known to be a technologist, futurist, and creative maverick, was talking about the implications of AI for creativity, ownership and the human experience, while also discussing his own new app FYI—an AI-powered messenger for creatives.
“We’re skirting a very thin line between authenticity, ownership,” he pointed out. “Do I own my facial map in facial recognition and AI? No. Does Beyonce own her timbre or her voice right now? No. We should. If Leonardo di Caprio’s voice can be mimicked and cloned to the T with AI, shouldn’t he own that?”
While all-powerful music labels come to terms with this new reality in courtrooms; artists like Dolly Parton are at war with AI companies. The line between “fair use” and “stealth” will be decided in legalese over the coming years. But tech companies, meanwhile, have been undertaking lofty projects built for headlines to paint AI in a beatific light—such as Chinese company Huawei’s questionable attempt to “finish Schubert’s Unfinished Symphony No 8” in 2019–with mixed results.
“These movements sound only a little like Schubert and a lot like film music,” wrote The Conversation in a review of the collaboration between a Huawei smartphone and film composer Lucas Cantor. “Allusions to Wagnerian harmonic suspensions and a clichéd orchestration do not make it easy to be otherwise.” Composers, concluded this writer, were under no threat from AI if this was what it had to show for itself.
Since then, AI has been able to do much more than a sketchy job of fill-in-the-blanks. In December 2023, Yamaha’s AI-powered piano, “Anybody’s Piano”, was in the spotlight at a Beethoven concert in Tokyo’s Suntory Hall—bringing Beethoven’s “Ode to Joy” to the fingertips of musicians with disabilities.
Yamaha’s piano and The Beatles’ experiment are both examples of how AI can be wielded for magical results. Around the same time, we were also introduced to Anna Indiana, the first ever all-AI singer-songwriter. In November, she dropped “Betrayed by this Town”—and was instantly dismissed as “deeply mediocre”.
Anna Indiana’s debut may not have been the smash hit that the world’s first real AI-pop song—2016’s “Daddy’s Car”, a collaboration between a system called FlowMachines and French composer Benoit Carre, supported by Sony—became. But her very existence is a sign of the times to come—terrifying for some, deeply encouraging for others.
9 AI tools for music
The loudest evangelists of generative AI in music talk of its transformative capabilities, of its potential as the great leveller in an industry that has been notoriously plagued by power plays, hierarchies, and gatekeeping. And indeed, this is true to some extent—if the rise of social media allowed us to re-imagine our creative selves, AI could change the process and make it still easier.
Musicians, content creators—or anyone who dreams of becoming one but can’t bring themselves to touch the ukelele they bought during the pandemic—can now have the same results for a fraction of what it would have cost to create a track in a studio, produce a music video and put it out into the world.
There are tools like Amper Music (acquired by Shutterstock), which can generate tracks using pre-recorded samples and allows users to modify music keys, tempo and instruments. Open AI’s Jukebox makes the job simpler—you can input information like genre, artist and lyrics, and Jukebox can create an original composition that is inspired by your prompts. Musenet, also owned by OpenAI, can generate songs with up to 10 different instruments in 15 styles.
Suno AI can help you “make a song” for any purpose or mood—including lyrics—based on a text prompt. This plug-in is now available to users of Microsoft’s AI chatbot Copilot.
Those who have more experience with composition and are more interested in the creative process can turn to AIVA, which allows users to create original music from scratch while also generating new interpretations of existing songs—for use in ads, video games or movies.
Beatoven allows you to choose between eight genres and 16 moods, and match the music to the tone and theme of your content—great for adding some original audio to reels. Producers can also turn to Orb Producer Suite to generate melodies, basslines, wavetable synthesiser sounds to create infinite musical patterns and loops for musicians.
If you want to generate, not necessarily create, music primarily for your own consumption, Brain.fm can whip up some tracks to make your own brain more focused. Ecrett can do the same based on scene and mood—such as party, travel, fashion, happy and serious.
The future is AI, in music or elsewhere—of that there is no doubt. “Of all the times to be alive, this one here is like wowzers,” said will.i.am on the Moonshots podcast in October. “We can get it right, and as an optimist, I think we will. Or we can get it very wrong.”
Which is to say, creativity and technology needn’t be strange bedfellows, if all is orchestrated well. “I’m preparing myself for this time when humanity has to be more human than we’ve ever been, because machines are going to outdo a lot of the shit that we used to do,” observed will.i.am. “And it’s going to force us to be more human to one another.” And what’s more human than the act of creating music?
Discover the latest Business News, Sensex, and Nifty updates. Obtain Personal Finance insights, tax queries, and expert opinions on Moneycontrol or download the Moneycontrol App to stay updated!
