By Alessandro Parodi, Olivier Sorgho and Matt Stock

GDANSK/LONDON (Reuters) – Inside a recording room at Queen Mary University of London, a group of researchers fiddle with novel artificial intelligence (AI) tools to develop what they call the “new virtual worlds” of music.

Andrea Martonelli and Max Graf are among more than 30 doctoral students working with Dr Mathieu Barthet, a senior lecturer in Digital Media, to explore computational creativity and generative AI. Together, they have set up a futuristic studio where music meets cutting-edge tech.

“It’s like extended reality, XR, is a way of extending the physical reality that we live in,” Graf told Reuters while showcasing “Netz”, his virtual instrument.

Netz is played through an augmented-reality headset that tracks gestures to create corresponding outputs, like notes or chords.

Martonelli plays the “HITar”, an advanced guitar with AI sensors, which reads his movements to make drum and synthesiser sounds. While the presence of AI in music-making can be traced back to the 1950s, recent groundbreaking advances in generative AI, with robots now making music as digital pop stars, have divided opinions in the industry.

Made popular last year by the ChatGPT language system, generative AI is capable of creating content including original sounds, lyrics or entire songs on its own, but artists often use simpler AI to enhance their sound. UK alternative rock singer-songwriter YUNGBLUD told Reuters he believes AI could help his music go “to another direction”. Other musicians worry that the technology could go too far. “I feel if you need AI to help me write a song, especially when it’s for a likeness, that’s not cool,” Amy Love from alternative rock duo Nova Twins said, referring to artist’s voices being artificially generated and adding that using dead artists’ voices is “not on”. In November, the Beatles released “Now and Then”, billed as their last song and featuring the voice of John Lennon extrapolated with AI from an old recording. Warner Music said in November it was partnering with the estate of deceased French singer Edith Piaf to re-create her voice using AI.

While labels and streaming companies partner to market the technology, many experts say AI raises legal and ethical concerns. “Unlawful development is what would put the sort of the opportunities of generative AI at risk,” said Abbas Lightwalla, director of global legal policy at the International Federation of the Phonographic Industry (IFPI). But regulation of generative AI is only in its early stages. “I think AI can have its place in the music production chain, again if it’s guided in the right way and if we ensure that the musicians keep a certain amount of control, and performers as well,” said Dr. Barthet. “But there might be situations where (AI) generated music works for new things that have not even emerged yet, new virtual worlds.”

(Reporting by Alessandro Parodi, Olivier Sorgho and Matthew Stock; additional reporting by Sarah Mills; Editing by Angus MacSwan)

Brought to you by