Discussing AI in Music

Below, my presentation for the AI conference in Folkestone on 19/11/23 part of the Folkestone Song Writing Festival 2023. I hope you enjoy it and if you would like to suggest answers to any of my questions below I would happily receive them.

Hello. I am Anna Braithwaite, composer, performer and community artist. I grew up in a village near here (Folkestone) called Nonington. I escaped to London at 18 where I studied fashion and then music. I have always been interested in cultural trends and predicting where we might go next, taking in everything I could from the arts, culture and news in order to be aware of, if not ahead of the zeitgeist. The word Zeitgeist is derived from the German words “Zeit” meaning time and “Geist” meaning spirit, both difficult to capture but exciting to attempt. Now working as a composer and experimental music creator I like to look at subjects in my work which create fear, to unpack them and expose their workings, to make them less scary. I have written about memory loss, being an outsider, death and change. I embrace change and have recently been working in VR and looking at new technologies which will allow fully immersive audio experiences. This is why, although I am fearful of AI and especially its effect on my industry and its workers, I was keen to be here today and delve into research about AI & music, to expose the subject and face my fear.

I am speaking to you as someone who knew almost nothing about the use of AI in music a few days ago but who has gained a general knowledge of the use of AI particularly in experimental music, my passion and have found, as with most things, the more I know the more interested I become.

I am dyslexic so I haven’t read any large tomes about the topic but I have listened to podcasts and watched TED talks, webinars and Tik Tok so that you don’t have to! By following my nose through the patchy information available, I have discovered a burgeoning, and curious world where humans tentatively sing duets with juvenile AI versions of their own voice, the AI doppelgängers of big music stars sing about their love of cats and AIs churn out what at best could be described as ‘cheap lift music’ to order. Fulfilling the insatiable desire, of people of the internet age, to have free music (and unchallenging entertainment), fast! AI’s reach, for now, seems to be at the far ends of the music industry, experimental and weird or mass produced and fluffy. The worlds of the singer song writers and pop artists seem safe at least. But maybe my other panellists will have something to say about that.

To be honest, I have just begun to scratch the surface with this subject and have more questions than answers, here are some of them:

Whose music are developers feeding to the AIs? Is that music (aka data) created by a diverse cross-section of humans? If they are simply feeding AI easily accessible, recorded music, how can this be representative of the diversity of music being made across the globe?

When the large majority of recordings from the past have been credited to white men, how can AI programmes avoid simply recreating more music that represents the tastes and concerns of this demographic? Women, people of the global majority, LGBTQIA+ and disabled music makers are not a niche, but their music is represented in only a small proportion of recording output (especially in the past), will AI continue to perpetuate this inequality?

When an AI ‘writes’ music in the style of a specific artist or genre does it then pick an ‘average’ musical response from its data set? If so, will this end up making it’s output all a bit average? I pressed a few buttons on the AI music generator site ‘Boomy’ and ‘created’ this: https://boomy.com/s/18081829. See what you think of it or have a go yourself. But watch out, if you upload to the internet and become an overnight sensation, Boomy own the copyright and even if you have completely changed the track in the editing process it won’t belong to you and you will only be able to collect 80% of revenue.

An image generated by Chat GPT to match the title of my Boomy AI generated track “Rainy day in Folkestone”

I work with people from the community and teach in schools. How will people achieve the confidence and wellbeing benefits music creation offers them if an AI does it for them?

I suggest AI is a nostalgia machine, scraping up and remoulding music from the past. How can it catch the zeitgeist, and like so many visionary human artists change our way of thinking by offering us a new vision of the future?

Don’t get me wrong, I am interested in technology. I am often an early adopter. I like software that can help me create things that I couldn’t do on my own, or helps me spend less time doing things I don’t enjoy. I use loopers to layer my vocals and instruments live, they allow me to create all the sounds in my mind, all at once. I use Sibelius to write scores more quickly and accurately and midi allows me to hear an admittedly naff but useful version of what I have written being played, but AI is not solving any problems for me currently.

There are problems with being a music maker that need solving, the biggest is that people love to hear music but they don’t want to pay the makers enough to allow them a decent standard of living. Creativity is not respected by society this is borne out by the invention of AI to write music. There are plenty of talented music writers, just in this room, not having enough people to write good music is not a problem people!

I look forward to a time when AI can help me make money doing what I love, until then I will continue to enjoy Eminem singing about cats.

Eminem AI cat rap: https://www.vice.com/en/article/88xadz/ai-generated-eminem-rap-youtube-chatgpt


Articles and talks that formed part of my research and may be of interest.

Some of the Artists who use AI:

Holly Herndon (Mat Dryhurst) - Holly +, Spawn, : https://www.hollyherndon.com

Jennifer Walsh: http://milker.org

Arca: https://arca1000000.com

Toro Y Moi: https://toroymoi.com

Gwilym Gold: http://gwilymgold.com/ Co-founder of London-based company Bronze. Its creators, musicians Lexx and Gwilym Gold and scientist Mick Grierson, hoped to create a piece of technology that would dislodge music from the static and fossilized nature of recordings. “We wanted a system for people to listen to music in the same state it existed in our hands — as a constant, evolving form,” They created Tender Metal album which exists only as an app in 2012. It doesn’t seem possible to access it anymore. Review of Tender Metal: https://www.theguardian.com/music/2012/sep/01/gwilym-gold-tender-metal-bronze  “It isn't possible to decide if it's good or bad, because "it" never exists in any fixed form.” “even if the 327th listen of Limbless turns out to be your clear favourite, there's currently no way of going back to it and ever replicating that enjoyment ever again.”

Anna Braithwaite