It was this spirit of playfulness and risk-taking that affirmed the extent to which all four of the Beatles became consummate musical professionals in the course of their eight-year career. So, what precisely were producers saving them from? The intersection between commerce, technology and culture has long been a place of anxiety and foreboding, just as how early sceptics claimed “records killed live music”. And yet, the creative apocalypse the industry has warned us about failed to arrive. Against all odds, the rising voice of artistes — varied, colourful and unmistakably their own — seems to ring louder than ever.
The debate on technology encroaching the mainstream consciousness of creators and consumers was renewed when Paul McCartney, more than 50 years after the global sensation he was part of broke up, recently announced that he was using artificial intelligence (AI) to create one last Beatles song, which will be released later this year. Will he take a sad song and make it better?
Alas, McCartney has not revealed the title or offered any clues about its lyrics. But speculation is rife that it may be Now and Then, a song Lennon composed and recorded as part of a demo tape in the late 1970s. After the Imagine singer was fatally shot outside his New York apartment building in December 1980, his widow Yoko Ono gave the old cassette to his band members, who were working on The Beatles Anthology, a career-retrospective documentary, record and book series.
See also: SoftBank’s Son ‘cried’ about Nvidia stake sale to fund AI bets qa
Lo-fi and embryonic, the tracks were predominantly recorded on a boombox as Lennon sat behind a piano in his apartment.
McCartney’s idea of releasing “new” material stemmed from the technological ingenuity of isolating the voice of Lennon from the aforementioned tape, which was used in an eight-hour epic documentary The Beatles: Get Back directed by Peter Jackson of The Lord of the Rings fame.
“[Jackson] was able to extract John’s voice from a ropy little bit of cassette where it had his voice and a piano,” said McCartney in an interview with BBC Radio 4. “We were able to take John’s voice and make it pure through AI and you were able to mix the record as you would normally do.”
See also: Bill edit url Gates says China is outspending the world on nuclear power url testttt
Realistically, the science is a tad more sophisticated than that. Emile de la Rey, the documentary’s dialogue editor, worked with Paris Smaragdis, a machine learning researcher at the University of Illinois Urbana-Champaign, to establish a neural network capable of separating individual voices or instruments and “resynthesise” them into clean audio. The AI then recreates the target voice by melding the information from the tape with the model developed from already isolated samples. To put it simply, you can tell the machine “these are vocals, and this is a guitar. Now lose the guitar”. The result of De la Rey’s collaboration was not an exact replica of Lennon’s voice but a close approximation of the real thing.
AI is a wonderful (or some say Pandora’s) box of curios that makes one excited about the possibilities but also fearful of its dangers and abuses. Canadian R&B singer and a long proponent of technological experimentation Grimes willingly helps artistes distribute songs using her AI voice if they agree to split royalties, yet former The Police frontman and bassist Sting is openly hostile about such crossovers. The proliferation of voice-emulating filters — methods that allow users to tweak existing vocals to mimic someone else, notably famous artistes like Drake, Michael Jackson and Taylor Swift — has ignited a new market for creative expression but also something more serious: a harbinger of headaches for an already muddied music industry before necessary copyright rules are in place.
Record labels and artistes are nevertheless sanguine about the future as they still trust loyal fans to tell apart a work or derivative from a deepfake, even if a bot can shrewdly replicate one’s musical tics. However, a dicey quandary looms: Are superstars, savvy they may be, comfortable with their pockets picked as trainable machines come for their art? Questions about ownership and authorship, coupled with knotty ethical concerns, linger as music generators cut into the vulnerable economy of working musicians. At the moment, protected intellectual property is only applicable to humans, but what happens when musicians, especially in the case of McCartney, collaborate with machines?
Finishing Lennon’s song with the aid of AI may recruit new listeners but it can also very well alienate older fans and purists because no one — including an optimistic McCartney visibly invigorated by a daunting challenge — knows what the late songwriter wanted to achieve. Others, who are more receptive and welcoming, consider the move a poignant tribute, as they recall the way McCartney used the same voice-separation technology to remaster the 2022 Revolver album and allow him to duet with his departed bandmate while touring last year.
The Beatles, known for beachcombing inspiration from headlines and news briefs and turning them into anthems that touch the private fragility of ordinary people, is a masterclass in the fraught art of making magic under intense pressure while breaking traditional norms. Judicious yet justifiably adventurous, the members knew exactly the amount of sparkles and sprinkles needed to produce songs that rattled the world. Perhaps the biggest takeaway from McCartney’s brave venture after all is this: It is not so much the use of newfangled tools to relive a memory, but their overuse that will undermine the musicality of one of the greatest bands in history.
