Listen to the 'final' Beatles song, 'Now and Then,' made with help from AI
The Beatles are back, sort of. The fab four just released a new song, the group’s first since 1995. “Now and Then” is being advertised as the final Beatles track, which makes sense given that two of the members have passed and the other two are well over 80 years old.
The song was built using a demo track from John Lennon dating back to the 1970s and a guitar track from George Harrison from 1995. The surviving Beatles, Paul McCartney and Ringo Starr, finished off the tune with the help of modern machine learning technology.
The software, which was first used by director Peter Jackson when making the Get Back documentary for Apple, was able to split Lennon’s vocal from the piano without any bleed, giving the remaining Beatles free rein to add whatever they wanted. You can watch a documentary on the making of the song right here.
As for the song itself, it’s a relatively slow Lennon ballad, with his vocals being a highlight. McCartney lays down one of his effortless bass lines and Starr hits the drums. The guitar solo is in the style of Harrison, but not actually played by him, though he handles some of the rhythm guitar in the background. There’s also a full orchestral arrangement because, well, it’s the Beatles. This likely won’t go down as one of the group’s most beloved tracks, but that Lennon vocal is both haunting and beautiful.
“Now and Then” was supposed to come out back in 1995 as part of The Beatles Anthology, along with “Free as a Bird” and “Real Love”, two other tracks culled from 1970s Lennon demos. Back in the 1990s, however, the technology just wasn’t there to separate Lennon’s vocal from the piano without some serious degradation to the original take.
The Beatles may say this is their last song ever, but we’ll report back in 2053 and see if that ends up being true. See you then. In the meantime, check out the track.
This article originally appeared on Engadget at https://www.engadget.com/listen-to-the-final-beatles-track-made-with-machine-learning-and-archival-recordings-153253946.html?src=rss