“I like music with a real sense of struggle; music that makes people feel, artistically, beyond uncomfortable,” says Scott Young. He means it. Scott native has spent years crafting his own blend of house and dub music, but with his latest release, A Model Within, the Hong Kong native truly takes listeners to the experimental fringes of sound.

Released under the alias ‘Satin Doll’, a moniker that he says grants him “much more freedom to create”, the album marks a striking shift not only in the style of music Scott is making, but how he makes it.

The EP’s opening track, Chants, floats between a church choir and the hum of a refrigerator, an open-ended reflection on texture and tone that never even recognises the need to establish a fixed identity. The secret to this chimaera of sound? The rapidly-developing technique of ‘timbre transfer’, the AI-powered capability to translate the sound of one instrument into any other instrument you can imagine (plus the ones you can’t).

Building on the approach to AI voice models employed by singers like Grimes and Holly Herndon, A Model Within uses timbre transfer not to imitate a single voice, but to present fascinating, new hybrids.

Similar to the raw, unrefined tones of the first sound amplification systems, the results are characteristically raw, temperamental, and full of sharp edges, yet with an undeniable influence that is already rippling out to the wider industry.

‍

Perhaps the most evocative example is Crytrumpet, which blends the cries of Scott’s own infant daughter with an AI model of a trumpet. Together, the two sounds create a sort of ‘meta-instrument’ – inherently human and emotive, whilst simultaneously metallic, resonant, and utterly uncanny.

Like many artists turning to AI, Scott’s initial interest grew out of the frustration he felt in his own creative process. “I was really tired of making music,” he admits. “I had put a few EPs out and I just felt fatigued. I didn’t want to repeat myself again, and I knew my work would benefit from finding new tools.”

“I knew my work would benefit from finding new tools.”

— Scott Young

In another life, Scott works as a software consultant. A recent project saw him working to create a series of audiovisual compositions for a client before this year’s AI boom. Rather than compose music for each individual piece, a manager on the project suggested taking a generative approach to the audio. It was a suggestion that sent Scott down his own rabbit hole.

Experimenting with a number of open source AI music models, he eventually found Neutone; one of the first ‘off the shelf’ audio tools using AI timbre transfer. Offering instrument models from harps and taiko drums to birdsong and distorted noise, the tech opened up a new world of possibility for Scott’s music. “I was continuously experimenting with new approaches, using multiple sound sources and observing the results,” he recalls.

Previously best known for his electronic music, Scott quickly references free jazz, Japanese noise music, death metal, and grindcore as his sonic touchstones. “I need all that music in my life,” he emphasises. “By contrast, I find the current music available in the market to be sometimes rigid; it’s well-composed but there is little chaos in it, little personal struggle.”

The unpredictability of the AI model’s output represented a welcome alternative, perfectly suited to the kind of music that Scott sought to create. “I want spontaneity to happen in the music,” he says. “Instead of being linear, each song on this album begins ‘in the middle’. There’s no beginning, no end, and you never know what’s going on. This is what I wanted to create.”

Scott acknowledges that the music on A Model Within may not suit more casual listeners, but stresses that he sees no problem with this. A large part of the history of music is a procession of niche sounds, originally seen as challenging, uncompromising, or even downright weird, that went on to reshape the mainstream, driven by their genuinely innovative approach.

“I see stagnation in the music industry.”

— Scott Young

Experimenters like Karlheinz Stockhausen and Dalia Derbyshire stretched boundaries of creative orthodoxy and, in doing so, paved the way for mainstream behemoths like electronic music, psychedelic rock, and synth-pop. The raw noise of The Sex Pistols and the New York Dolls was disdained by industry elites before being coveted by the masses. True innovation happens on the margins.

“This music requires a certain level of attention and critical listening,” he admits. “I feel there’s no need to push the music towards more consumer-based audiences; they already have an abundant supply of music. This is made purely from a sonic perspective, thinking about what will induce changes to the listener’s psyche.”

While Scott may currently be in the early minority of musicians using generative AI in their work, it’s a creative community likely to grow exponentially over the coming years, given the rapidly-increasingly simplicity of the tech. In Scott’s opinion, all that matters is giving people a first taste of what AI can do.

“People need to have their first forays into AI,” he assures me. “Perhaps initially that’s using prompts to test if they can generate the simple sound of a ‘clap’, but I definitely see more people getting into timber transfer.”

While some look ahead with trepidation to that future, Scott anticipates an opportunity to break out of the increasingly tired styles that dominate the industry.

“In music these days, I see ideas being recycled due to a lack of new musical movements,” he observes. “I see stagnation in the music industry. People are using the same old tropes to articulate their story, but new technologies can change the status quo and allow us to progress a little bit further.”

Amongst those changes, Scott hopes that AI could prompt shift away from personal narratives and cults of celebrity, and towards a deeper appreciation of sound for its own sake. “People often relate to music because they are influenced by the backstory behind the music. I believe we can focus more on the timbre of the sound and its ability to trigger emotional responses, rather than the backstory.”

While A Model Within may have been Scott’s starting point with generative AI, he makes clear that his journey into this creative field is only beginning. Over the longer term, he’s looking forward to a time when more complex sounds and instruments can be accurately modelled not just by teams of researchers, but by independent creators.

“Having my own model is the next goal,” he explains, noting that achieving this in a way which reflects the enormous range of possible pitches and dynamics is difficult. “I’ve been training with various materials, such as piano, drums, but it’s still not quite there.”

But any worry is nullified by technological progress that Scott sees as inevitable. VR simulations and Neuralink-style brain-computer interfaces complement AI music all in the same breath. “All this technology is still in the trial and error stage,” he smiles. “Really, we're just getting started.”

“New technologies can change the status quo and allow us to progress a bit further.”

— Scott Young

‍