AI Helps Make a Beatles ‘Last Record’
Beatles’ cofounder Paul McCartney said this week that artificial intelligence (AI) has recaptured the voice of former Beatles member John Lennon from old recordings. Lennon was murdered in 1980.
Music that includes Lennon’s voice from the past will become what McCartney calls “the last Beatles record.”
The 80 year-old artist spoke to the BBC. He said the AI technology was used to separate the band members’ voices from background sounds in an old recording. Film director Peter Jackson was making a documentary series about the band at the time. The series, The Beatles: Get Back, came out in 2021.
McCartney told BBC radio that Jackson was able to separate Lennon’s voice from the musical sounds in the recording. “He could separate them with AI, he’d tell the machine ‘That’s a voice, this is a guitar, lose the guitar’.”
He added, “We were able to take John’s voice and get it pure through this AI so then we could mix the record as you would do.
FILE - Paul McCartney poses for photographers at a film premiere in London, December 12, 2022. (Photo by Scott Garfitt/Invision/AP, File)
McCartney did not identify the name of the demo song that Jackson used for the process. But the BBC and others said it was likely to be an unfinished 1978 love song by Lennon called Now and Then. The demo was included on a cassette with the words “For Paul.” The BBC reported that McCartney had gotten the cassette from Lennon's wife, Yoko Ono.
McCartney described AI technology as “kind of scary but exciting.”
Last year, the same technology permitted McCartney to sing a virtual duet with Lennon. The performance took place at the Glastonbury Festival in Britain.
FILE - Singer John Lennon appears during a press conference at the Hotel Americana on May 13, 1968, in New York. (AP Photo, File)
Holly Herndon is an artist with a doctorate in composition from Stanford University in California. She used AI machine technology on her last album, 2019's Proto. She suggested that the Beatles' recording was likely created using a process called “source separation.”
"Source separation has become easier to do with machine learning,” Herndon said. The process lets a user take a voice from a recording and isolate it so that it can be paired with new music, she explained.
That differs from a deepfake vocal. “A deepfake is an entirely new vocal line,” she said. It comes from a machine learning model trained on old vocal lines, Herndon said.
McCartney is set to open a show later this month at the National Portrait Gallery in London of his private photographs. He took them took during the early days of the Beatles at the start of “Beatlemania,” when the band rose to worldwide fame.
The show is called Eyes of the Storm. It includes more than 250 photos McCartney took between 1963 and 1964. They include pictures of Ringo Starr, George Harrison and Lennon, as well as Beatles manager Brian Epstein.
Words in This Story
demo –n. a recording intended to show off a song or performer to a record producer
cassette –n. a container holding film, photographic plates, or magnetic tape
scary –adj. causing fright
exciting –adj. producing excitement
virtual –adj. being on or simulated through a computer or computer network
duet –n. a composition for two performers
composition –n. the writing of music
https://learningenglish.voanews.com/a/ai-helps-make-a-beatles-last-record-/7138954.html
