Skip to content

The Beatles Release Their Last Song Thanks to AI

How AI assists in recreating the voice of those musicians who passed away.

Photo by Fedor / Unsplash

The Beatles' last song, Now and Then, on which John Lennon, Paul McCartney, Ringo Starr, and George Harrison worked together, is finally released on November 2.

And that's thanks to the same artificial intelligence tech that was used by Peter Jackson in his 2021 The Beatles: Get Back docuseries. To 'create' the song, Wingnut Films artificial intelligence was exploited to isolate vocals from Lennon's old demo, included on a cassette among other demos that Yoko Ono gave McCartney after Lennon was killed.

The track also features electric and acoustic guitar recordings from 1995 by the late George Harrison, backing vocals of Paul McCartney and Ringo Starr, Starr's drums, as well as bass, piano, and a slide guitar solo by Paul McCartney.

The Now and Then song was originally recorded in the late 70s by John Lennon after the Beatles broke up. Now, the song is already available for pre-save and pre-order.

Fans can also watch a 12-minute film about this 'new' recording that will be released next Wednesday. The music video for Now and Then will be released on Friday, November 3.

This is not the first time AI has been used to recreate the voice of those who passed away and allows fans to hear their favourite musicians sing one last time. Back in 2021, Supertone ‘resurrected’ the voice of the late South Korean folk star Kim Kwang-seok. In November 2022, Tencent Music Entertainment used voice synthesis technology to create voices of several late artists, including Teresa Teng and Anita Mui.

An artist in Costa Rica has used AI to record a duet with his late father, the legendary Costa Rica rock singer José Capmany, and the track has been released on Spotify. And in 2021, after the punk cabaret band World Inferno lost their frontman, Jack Terricloth, they were able to extract Terricloth's vocals from the band's rehearsal recordings with AI and create a finished song that sounded like Terricloth had recorded it himself.

As AI technology advances, we're likely to see more of such examples. If you want to read more coverage on AI in music, check the Kill the DJ's AI section.