The music industry is facing a revolution called artificial intelligence. Generative AI models that create terrific fake songs—like Drake’s and The Weeknd’s—are one side of that coin, but there’s another. Which, for example, allows the Beatles to release a new song this year, more than half a century after their last recording together.
Paul MCCARTNEY. The Beatles member stated on BBC that he had used an artificial intelligence system to “extract” Lennon’s voice from an old demo to complete a song by the group that will be released this year, although McCartney did not give specific dates.
All thanks to a tape. The song is expected to be “Now And Then,” a song written in 1978 by John Lennon. It is one of a series of songs that Lennon recorded on a boom box in his New York apartment, resulting in a tape titled “For Paul.”
an old acquaintance. The song could have been included as part of a Beatles “reunion song” earlier in 1995, because at that time the group members were reuniting their Anthology. Yoko Ono, John Lennon’s widowsent the tape to McCartney at the time.
tried before. Two of the songs on that tape, ‘Free as a bird’ and ‘Real love’, were remastered by producer Jeff Lynne, becoming the Beatles’ first “new material” in 25 years. The song ‘Now and then’ was also tried to recover, but although it had a chorus “it almost completely lacked verses”, Lynne explained then. “We did the backing track, a rough attempt that we didn’t really finish.”
So can not be. McCartney stated in the BBC interview that George Harrison refused to work on that song because the quality of Lennon’s voice was “rubbish”. For him “it didn’t have a very good title, it needed some work, but it had a lovely chorus and John sang on it. But George didn’t like it, and because the Beatles were a democracy, we didn’t put it out.”
The idea. According to the artist’s statements, the idea came after the documentary ‘Get Back’ that Peter Jackson premiered in 2021 and that narrated the production of the album ‘Let It Be’. In that project Jackson used a computer system to be able to separate and extract the voices from the ambient noise and from his own instruments to obtain the clearest and cleanest audio possible.
John Lennon’s voice. Jackson, McCartney claimed, was able to extract John’s voice from a poor-quality cassette. “We had John’s voice and a piano and he could separate them with AI,” the artist explained. The film director and his team simply helped to label and distinguish the sound of Lennon’s voice from, say, his guitar, and from there they could extract the voice of the legendary Lennon.
to remix. Thanks to this technology it was possible to separate Lennon’s voice to remix it as it has been done for years, already including the rest of the necessary instruments and voices. What has been achieved is therefore to record the original song, but to do it with much higher quality and without losing Lennon’s original voice in it.
Beware, “this is really Lennon”. McCartney was amazed at what this technology had made possible, but also warned of the dangers it poses, for example to recreate Lennon’s voice singing other songs. “That’s just AI”, and unlike that possible future, in this song “this is really Lennon”.
Image | Wikimedia
In Xataka | The shock of an era: this was the murky breakup of the Beatles seen 50 years later