How Artificial Intelligence is changing the music industry and whether it will help

The Beatles are the latest band to release a single with the help of AI Credit: PA Images

The Beatles have announced their final single, written and performed by the late John Lennon.

Along with the announcement came a clip featuring an image of a white cassette tape projected onto iconic Beatles' locations such as Penny Lane, Strawberry Fields, and John Lennon's childhood home in Woolton.

But how could the band produce a track when half of its members are no longer alive? Instead of finding replacements for lead guitarist George Harrison and frontman John Lennon, the band is using Artificial Intelligence (AI) to restore old demos and create their final single.

The Beatles at a 1960s recording session in London Credit: PA

How is Artificial Intelligence being used in the music industry?

With the recent advancement of AI, musicians and producers are already taking advantage of new tools to create original music and clean up or enhance existing audio.

Even streaming services like Spotify use AI to help recommend songs and artists to their users.

When Paul McCartney first announced AI had been used to create the final Beatles single in June 2023, fans began to worry the John Lennon's voice had been artificially recreated for the track.

However, the 80-year-old assured fans on X, formerly known as Twitter, that "nothing has been artificially or synthetically created".

Instead, AI tools were used to clean up original demos in a process called stem separation.

What is stem separation?

AI may have the potential to create original songs, but it is currently mostly used to restore and clean up existing music.

Stem separation is a process which makes it possible to separate instruments and vocals from a single track to isolate voices or certain parts.

It has been common practice for decades to record each instrument and voice in a track separately, as it allows for easier mixing and the potential for future remixes.

However, this final Beatles single features singing from John Lennon, which was recorded in his New York apartment shortly before he died in 1980.

The band first tried to record a track featuring these vocals in 1995, but Paul McCartney said the piano John Lennon was playing in the recording made it impossible to sample.

George Harrison's Widow, Olivia, said: "George felt the technical issues with the demo were insurmountable and concluded it was not possible to finish the track to a high enough standard"

Now, using stem separation, The Beatles are able to take John Lennon's vocals from the recording while also removing the piano and any other background noise.

It's unlikely that AI was only used for removing the piano from the recording, as there was also reportedly lots of background noise.

How can AI help clean up audio?

While stem separation is a revolutionary tool capable of bringing life to old tracks, AI is also able to help streamline and simplify existing technology.

When it comes to improving audio quality, companies like Adobe and Avid have created hundreds of methods to remove background noise, isolate sound, or remove any echoes.

For The Beatles, there were technical issues with John Lennon's original recordings, due to some persistent “buzz” from the electrical circuits in his apartment.

While tools to fix issues like this have been available for years, AI can automate the process and turn what would be hours of fine-tuning into a simple press of the button.

This makes audio restoration far more accessible for musicians without the technical knowledge required to use these advanced editing techniques.

Can AI produce a full song?

While The Beatles may not be using AI to create music or write lyrics, others are already using the technology to create original pieces of work.

In a viral Twitter post in June 2023, a Liverpool fan asked ChatGPT to write a celebration chant for Alexis Mac Allister, who had just joined the club.

The AI tool was asked to "come up with a song for Liverpool fans to sing about Alexis Mac Allister (to the tune of a 90s dance classic)."

ChatGPT produced a football chant to the tune of Ecuador by Sash.

While it may not be heard in Anfield any time soon, it is clear that AI is capable of taking highly specific instructions and using them to produce relevant lyrics.

Who else has used AI in their music?

Smaller artists have already used AI to produce music and vocals for entire albums, but an entirely AI-generated song has yet to see mainstream popularity.

In April 2023, a TikTok user released a song called 'Heart On My Sleeve' which featured vocals from The Weeknd and Drake produced entirely by AI.

The song was quickly removed from social media websites after a take down notice from both of the musicians.

Megan Steinberg, a PhD student at The Royal Northern College of Music in Manchester, has been using AI in her compositions.

In her recent piece, De Gipsie Moon, she fed the beginning of a piece by Erik Satie into a generative AI that predicted how it should end. The AI's prediction was then played by an ensemble.

She said: "It is a very interesting, complex and wild motif very different to the original."

Through her work, Megan hopes she can show the uses and limitations of AI as a collaborative tool.

She said: "I think it’s just more powerful way rather than like lecturing people... I’m still figuring out how music and art can be used to convey these messages but I think people need to have these conversations."

Can AI infringe on copyright?

It is unclear whether tracks like Heart On My Sleeve infringe on copyright law, however the UK Government proposed in 2022 granting copyright exemptions to AI companies for the use of training.

This would mean that books and music could be used to train AI's without the consent of the copyright holder.

However, the exemption was criticised by MPs and musicians and is now being reconsidered.

UK Music Chief Executive Jamie Njoku-Goodwin described the proposed exemption as a "green light to music laundering."

Currently, the Government allows an exemption for text and data mining for non-commercial research purposes and allows creators to licence their work for any further purpose.

Want more on the issues affecting the North? Our podcast, From the North answers the questions that matter to our region.