Holly Herndon gives us the rundown of her tech-filled composition career and the future of AI and machine learning in music.
You’d be hard pressed to find an artist like Holly Herndon, her unique technology-based approach to music composition sets her apart from the pack. This intersection between AI, technology and music creates an area of intrigue for the artist, a cul-de-sac where she and partner Mat Dryhurst keep learning about and keep creating within.
Holly notes that the world of laptops and technology-based music came after moving to California and attending Mills College, a renowned liberal arts school. “That’s when I encountered so many DIY technologists, it was just a really great space to learn about using the computer as an instrument, because it was a very nurturing environment,” she says.
“That’s also when I learned how to use visual programming languages in an environment that encouraged performance and fast iteration, but it wasn’t about trying to build the perfect instrument, it was more about trying to build something you could make art with.”
Keep up with the latest music news, festivals, interviews and reviews here.
A lot of Holly’s work, along with Dryhurst, has been centred around the at times complex realm of AI and machine learning and the endless possibilities of what it can create.
Often in the music world, there are discussions of the high barrier to entry in the world of machine learning and music. Herndon believes it has a lot to do with the language surrounding it. “I think all instruments have a barrier to entry in some way,” she continues, “it varies.”
“The laptop and what machine learning can enable is interesting, because yes, you have to develop your own systems and understand the inner workings of it, it is quite complicated and does require spending a lot of time understanding it.
“But at the same time, it has the potential to enable people who might not otherwise be able to create works the same way. For example, the work I’m doing right now is prompt based, especially in the visual world.
“I’m able to animate entire sequences by using text prompts of what I want to animate, and I’m not an animator. If I tried to go in and hand draw in Blender or whatever, I never learned how to do that. So, it’s kind of both, lowering the barrier for entry in understanding and developing.
Holly, along with Dryhurst is perhaps best known for the creation of her AI Baby, SPAWN, a singer that recreates certain audio elements in Herndon’s voice.
There will be some reference of this baby in Herndon’s Saturday talk, which she hopes can help music practitioners understand the confusing elements of AI and machine learning, while clearing up the sticky areas of these fields.
“I hope that I can demystify a little bit of ML. I feel like when people talk about it, it’s often in this grand, futuristic kind of way, I want to talk about it in a today kind of way. So hopefully people can get it – maybe it’s demystified a little bit – and then they can be interested to pursue their own research.
“It won’t be a dry discussion, I do like to show some examples of ML in action. I think I want people to think ‘How will this affect my music practice?’ then that can lead into a conversation about what it means. Another thing I’m talking about right now is identity play: this idea that you can map the timbral quality of one instrument onto the performance of another, which opens up a possibility to perform through one another.”
“If they were willing, I can perform through someone’s voice, and they can perform through my voice, and that’s where it starts to get really weird, but it’s really interesting to talk about, and can unlock a lot of cool work.” It’s clear that the show is set to be an extremely informative one.
It takes place at RMIT’s The Capitol this Saturday, December 11, tickets still remain, which you can grab here.