Hello and welcome to the Sound On Sound Podcast about Electronic Music and all things synth. I'm Caro C and in this episode I'm talking with Vicky Clarke, a sound and electronic media artist from Manchester, England, working with Sound Sculpture, DIY, electronics and Human Machine Systems. Vicky creates electronic artworks that explore our relationship to technology. Vicky won the Oram Award in 2020 for Innovation in Sound. She's an In-Motion composer for Sound and Music UK for 2024 to 2025, developing her work in spatialised sound for a new installation called Latent Spaces, which you'll hear more about imminently. Vicky produces music under the moniker Soname and her debut album, Sleep States, was released in 2022, accompanied by net art piece sleepstates.net. Since 2019, Vicky has developed a specialism in machine learning and musique concrete through an artist residency with Novas at the University of Manchester, in collaboration with Prism at the Royal Northern College of Music. As is customary, we start with the taste of Vicky, a.k.a Sonam's music. This is an extract of a track called Sleep States from the aforementioned debut album. This opens the doorway to lucidity. The method of projection is dependent on the device you use, your ambition and your budget. Hello Vicky Clark and welcome to the Sound On Sound podcast. Hi Caro, nice to be here. Great stuff. So yeah, we've got lots to unpack. Fellow Manchester creative, technical, electronic music sound person, so that'll be fun unpacking your more recent work especially. Yeah. So I'd kind of like to start with a bit of a potted history if you like, but really what I'm interested in is how did you get to the point of fascination you're at now, what were the main things that feel like led you to the point where you're at now in terms of your work? I do love working with technology and electronics. I guess, I grew up in South Manchester, my dad was a computer programmer, so in the 80s we always had bits of computers lying around the house, bits of technology, but I wasn't so interested in that sort of stuff then, it's just in later life, these kind of materials have come back to me, I expect. But yeah, in terms of sound, I was very musical as a child, very traditional instrumentation, trumpets, guitars, that kind of thing. But it was actually through field recording that I got into thinking about recorded sound and that manipulation of material. I just started out making field recordings. I did a short course at Spirit Studios in like the early, oh my God, 2010s, maybe called Wise Amp, which was Women in Sound Engineering and Music Production. Looking back on that, that was probably my sort of way into working with kind of digital audio and DAWs, so introduction to Ableton and Logic, that kind of thing. And I just found it fascinating that you could, yeah, take a sound recording, put it into a computer and then manipulate that electronically, and that really opened up this world of, yeah, sound as a material the way that you engage with the world. So there was that kind of field recording side and that kind of production aspect. And then separate to that, I used to work in youth arts for a long time and I quit my job and went to art school for a year, did my art foundation as a mature student and at that time, I sort of got into, I was in the metal room a lot of the time working with these kind of sculptural forms, lots of materials, and I started to mic up these materials and listening to the sounds within them, so using contact mics and I started to read about musique, concrete and the sort of history of that. So kind of working with recorded sound, manipulating that, listening into material and was something that I found really fascinating and kind of brought that kind of, um, music concrete process with my field recording, uh, interest into kind of a more, um, solo practice. Then I guess there's, with the electronics, um, around that time as well. Um, I started a project called Noise Orchestra with my friend Dave b, um, who's a, an improviser, and we sort of taught ourselves to build little oscillator circuits. Um, so really DIY learning things off the internet on different residencies. Um, and yeah, I became really fascinated about this idea of working really hands on with. Why is it electricity? This physicality, we sort of had this about five years. We sort of made a lot of different, um, yeah, oscillators and circuits that kind of used light and sound. We did installation work, residencies and it was through that that I sort of got more into building my own interfaces. Um, and just sort of brought that more DIY electronics practice into my kind of more solo music production work. Yeah. So those kind of grounding points of field recordings, DIY electronics, thinking about more kind of sculpture and sound, and those are the kind of threads of my kind of solo work now. That's how I sort of, they're the materials I used to perform with and what I kind of, yeah, what I do in my studio, essentially. Wonderful. I believe it also involved a trip to Russia. Oh, it did, yes. Back in the day. Yes. So in 2015, uh, Dave and I, we were working with the National Science and Media Museum in Bradford. Um, so it was inspired by the Russian avant-garde of the 1920s working with graphical sound and these noise machines. So yeah, as part of that residency we got to travel to Moscow, um, and meet the director of the Thein Center, Andre Smirnoff, him, who wrote the book, sounding Z. So yeah, we got to play on these, um, amazing original thein machines. The Rhythmicon, which is one of the first drum machines. It was, it was amazing. So I've always had that love for thinking about, um, kind of media archeology, thinking about old, old technologies, how they impact on the future and how we're kind of just recycling some of these, these things. But yeah, that love of hands-on, yeah. Electronics is kind of see me through. Cool. And so how's that all led you to neural synthesis? So, yeah, I guess bringing that, um, that kind of very material practice. So working with, you know, amping up metal, um, thinking about recordings, working with drum machines and samplers. I'm really interested in working with sound as a material that it can be manipulated and kind of take you into new worlds. And in about 2019, I went on a British council trip to Russia to sort of find out more about this. Um, well I came across this technique called neural synthesis. And I heard the kind of output materials from this computer model and I was really quite wowed about this. And I sort of thought, how do these new neural networks and parasynthesis models, how can I think about those in terms of my own practice? How could I think about, um, the effects that they may have on the materiality of sound and work with these new systems, which were quite emergent technology at that time? Mm-hmm. So when I came back to England. Um, in Manchester, I found that at, uh, RNCM, the Northern College of Music up the road, they just released a model called Prism Sample, RNN, which is an example of a neural synthesis model. So I got in touch. Um, I ended up doing a residency with the University of Manchester and RNCM for a couple of years exploring this kind of new sonic turn in neuro synthesis. Um, so thinking about machine learning and music concrete and yeah, just exploring those. Those two fields really, and seeing where I could get to. So may I ask a bit of maybe an obvious question coming from very sampling background, kind of what's the difference between sampling and then the neural synthesis? Am I right that you are feeding in sounds? Yes. Yeah, exactly. So neuro synthesis, it's kind of, um, yeah, on a really sort of basic level, it's just a way of synthesizing audio, right? I do see it like that. It uses neural networks to synthesize audio. The kind of components of it are that you have to work with your. Datasets of sound. So what that means is field recordings. So I work very much, well, I only work with my own materials, so I was recording sounds of, um, kind of, uh, industrial heritage. So I had, um, a dataset called the Aura Machine Dataset. Um, and that included sounds of electricity, sounds of glass, metal and wood, and also the sounds of mill spaces. And when you say sounds. You also mean electromagnetic recordings, don't you? Yes. Yeah, definitely. So there was, I wanted to have a data set, um, so a kind of, um, a corpus of material, so lots of field recordings that had these very distinct classes of sound. So if you had, um, yeah. Field recordings of wood and glass of metal and electricity. Um, if you process these within your own network, um, what would be the output of that? Would you get a kind of new materiality or would you kind of get a kind of pale limitation of the input? So yes, a parasynthesis it take. Inputs of field recordings. It has something that's called a, a chunking script. And what that is, it just kind of, um, basically makes very small samples of your field recording. So like six seconds. And then that sound becomes data that is then fed into the machine, in, into the model. Um, and then the model, the parasynthesis model trains on that data. So the sound is now. Just data points, statistics, um, and then it tries to detect patterns and features within that material. Um, yeah. And so every time it trains, um, through that whole data set of sounds, that's called an epoch. And you can take a sonic output at the end of each epoch, so it kind of trains over and over. So this is called the training or the learning stage. Um, and then those outputs, uh, earlier epoch sound very noisy. Because it hasn't learnt, it hasn't trained on the material to kind of determine the sounds in there. And later epoch you can start to hear more familiar sounds. So the sounds of glass were coming out, some of the metal sounds and then these kind of quite strange combinations of materials or strange architectures. Um, and. Kind of rhythms were coming out in tones. So for me it was quite interesting because it sounded similar to some of the input recordings that I was feeding it, but also very uncanny and very otherworldly. So that was what I was trying to get to, this kind of idea of could these systems produce materials that I couldn't make as a producer, and how might I composed with those? Um, what do those feel like and sound like, and how did that work out? What did, what did they feel like and what did you compose? Yeah, so this was, so this is now about four years ago actually. So the first piece I made was called Aura Machine, and the idea of that was, it was sort of a challenge, can a machine. Generate an aura. Um, can these sounds be beautiful? Can they have their own character or are they just pale limitations of what I've put in as the human? And I was very interested in where is the hand of the artist in these systems and what is this human machine collaboration? And yeah, I found that, um, I was really surprised actually. So I was. It was during lockdown, I was listening back to some of these output sounds and it was a very uncanny space to inhabit as a composer listening to these materials that were so familiar yet different. Mm-hmm. Um, 'cause I think we're used to always knowing as record is where our sound sources are from. We have this relationship to our material. Mm-hmm. But this time it was, it was our material, but slightly, yeah. To one side. Um, and I found that very interesting. So yeah. And I did find beauty within. Within these materials. So for example, there was one of the first sounds that I kind of detected was this kind of combination of glass and static, which was this very kind of, um, yeah, ethereal sound that I couldn't, as a producer make myself, but it was still me in that material. So it was those kind of relationship. And those sorts of sounds that I found really interesting, um, to work with. So yeah, so the challenge then was, was to think, how can I yeah, work with these materials? They're very noisy when they're kind of output. Um, so I kind of started to classify those in terms of tones and events, um, different feels and moods, um, to then compose the piece. Um, so that piece was called Aura Machine. And the idea was that it would take the listener on a journey through training a neural network. So it would start off with the raw field recordings. It would then move to kind of a speculative state, which is kind of the latent space. This kind of like transmutation space where this material is in flux, where the models training, and then the end, end of the. The composition was the output material. So just those kind of more machine learning sounds. Um, and it was a very lo-fi piece 'cause I was working with models from like 20, it's 2019 sample, RNN. So it's a very early model of neuro synthesis. So it was a very, it was like 16 K some of the audio, very lo-fi. And now from kind of our 2025 standpoint, looking back, it seems that, that that model and those sounds of that time already kind of. Dated in a way they're kind of retro. It had its own tmba, almost like tape did or like other formats did. Yeah. Wow. Exactly. Carro and, yeah, and the tape is a really good kind of analogy because, um. Yeah, I was interested in the sort of connections between this and music concrete, which worked with tape. Mm. And for me, some of those sounds that were coming out of this model, they actually sounded like early tape experiments with this noise. Um, and I lo, I love noise Caro, as you know, as you know. Um, so yeah, so it kind of, there was a real, um, aesthetic connection there to me as well in terms of the actual material. Wow. And so how much has that evolved in the space of five, six years? We are living now in a kind of generative AI hype curve. Mm-hmm. I guess you'd call it. Totally. Totally. We were living in kind of an AI doom world as well. Obviously it's an emerge an emergent technology. Technology moves on and advances so quickly. I guess now, yeah. Looking back to that time, even though it was just five years ago, it did seem kind of more primitive at at that time. Yeah. I guess the history of AI is just interesting to look at as well, because there are these, they're called AI Summers and winters, so there's always been these. Yeah, these hype curves where there's more interest from the public or, um, the technology takes a leap forward. Um, I guess the thing that's different about this current curve that we're in is that, um, with deep learning models, they're having much more of a, of an impact. Um, whether that's environmentally, you know, there's concerns around artistic. Um, data sets and copyright all of this. So, and all of these concerns are coming to the forefront in the media, which are really important. I think it's the profit side of it as well as all the sort of the business side of it, I think is a big part of it, isn't it? It is, yeah. And that's, that's really crucial car, because the point is, is about who owns, who owns the power and control of these systems. Yeah. And I'm all about the DIY, um, and the sort of collective. So for me, I. Sort of ethically, I have to work ethically with these technologies. Um, so I work in a very local and lo-fi manner. Um, you know, these, for this piece that I'm making at the moment, I'm reusing materials from old training. Um, I'm not doing any new training. I don't use data centers. It's all trained locally. Um, and all the material is my own recordings, so I'm not harvesting or scraping any data. Um, so there are ways to kind of, yeah, work with these technologies in more kind of new. Nuanced ways that are about kind of low-fi and DIY roots. Yeah. And just recently, I think at some of the major festivals as well as some motech this year, um, they've had a Rewilding AI focus. There's, um, also a festival in Amsterdam called the Slow AI Festival, which is around changing our approaches to some of these technologies. So. I think this is always the case with emergent technologies. There's this, this, the kind of evoke hope and fear at, at the same time. Um, and we have to find our own ways to, to work with them or, or not. Okay. One thing I'm thinking, which might be, um, again, back to basics is the. When you talk about a model, is that, are we talking hardware? Are we talking software? Can you tell me a bit more about that? Yeah, sure. So I guess with machine learning and anything in kind of data science, um, you're working with different models, so different programs, I like to think of them as programs essentially. Um, so the model, um, that I use was called sample, RNN, um, prism Sample RNN. 'cause it was from the RNCM. Um, and this is an open source model, so these, you can just. Google them, find them on the internet and use them yourself. This exists currently online under a Google CoLab. So it's like a coding, um, almost like a door for coding essentially. And you can actually input sounds in that way, just, just from your laptop. It's all open source. So I worked in collaboration with a creative technologist called Dr. Christopher Malin at the university. So those models you can kind of. Upload your, your samples to, and you could, you could train them from your laptop using the cloud. Um, so that would involve kind of training. Yeah. Within the server. Um, but I actually did my training at RNCM all locally on a, on a desktop, on a, on a machine. Um, so that's where you're sort of getting into that gray area of, we talked before about, you know, these large language models that are kind of using a lot of water, for example. That's because when you ask chat GPTA question, um, it's going to a data center. Um, and it's using a lot of energy to kind of answer that question. So I. Don't do that. Um, I very much train locally on a, on a desktop computer, so it's kind of using less energy in that respect. And if you're doing it online via the Google CoLab. Using, you're back to using the, a lot of water approach you are, I think you can actually calculate how much energy you, you are using. Right? So this is why it's interesting to think about where these models are hosted. Right. Um, and what we can do to make these more locally, um, available. Mm-hmm. Yeah, definitely. So what is it that you are actually wanting, or let's say with the latest iteration of, of this work, um, what is it you were actually wanting to compose? What was your kind of. What's your kind of message as an artist with those sounds, if you know what I mean? Yeah, definitely. So I'm working on a new, a new installation piece, which is gonna be a spatial sound piece. It's called Latent Spaces. Um, and the idea is that you can step inside a computational model and experience this latent space, this kind of data dimension where materiality is in flux. And, and what I want to kind of, yeah, sort of provoking people is the idea that we're surrounded by these invisible. Technological systems that really impact our day-to-day lives, but what do we really know about them? And so when I was doing my research on, on the, these projects, I became really fascinated, um, by the idea of kind of hidden and invisible technological systems. And they're talked about in a very sort of, though they're at once statistical and they're just to do with code. They're also. These spaces that, that people don't really understand. They have these multiple layers. They're kind of imaginary spaces. So they, they have this kind of dual kind of purpose or function as both mythical spaces, spaces of alchemy almost, but also just kind of statistical spaces as well. So I was interested in this idea that, um. Maybe we could step inside these spaces and actually feel and hear what this latent space is like. So I wanted to kind of, yeah, I thought that would make a good premise for an installation piece where people could step inside, um, and actually hear these materials. So yeah, so that's the idea. So it's a sound installation and it would be working with, um, sculpture and audio. Um, and you'll hear. These sounds that are kind of, um, sounds of Manchester's industrial past, but through these kind of neural networks. So there's recordings of Cory Bank Mill machinery, so kind of quite rhythmic material in there. Um, you'll hear these kind of materialities of glass and steel and metal, um, and actually hear the kind of Tumblr of these models as well. I wanted people to be able to hear what these systems sound like, what does the machine sound like? 'cause they have their own character as well. So that's the idea. It's kind of, yeah, being able to step inside these spaces so we can kind of maybe understand them a bit better. So it's kind of creating it that interdependence, that intimacy, if you like. And is that partly to. Bust myths or to, to sort of bring people away from just being consumers of these things? Yeah, I think it's playing with that mythology, actually. I think when emergent technologies come out, there is all of this hope and fear around these technologies. So, you know, you could be really literal and say, this is just a statistical model. This is doing this. But actually the way that. The people that write the code, even in, in these textbooks, are describing these models. They use language such as, um, unknown spaces or hidden layers. So it is also very mysterious in how the actual technologists describe it. Mm-hmm. Um, and one thing actually, just in terms of a visual that I became really fascinated in was in these kind of machine learning textbooks, they kind of depict 3D space. Um, and it's always these kind of, yeah. These square, these 3D square boxes, these kind of nets with all these data points inside and this data like swimming around and all these colors. Um, and then also these kind of like architectures of neural networks are very kind of, um, yeah, it's hard to describe in terms of, in verbally. Um, but yeah, they're very sort of mysterious drawings. And so I became interested in the idea, could you step inside these 3D boxes? So that was kind of a. Kind of like a visual cue for the installation as well, to be able to step inside. Yeah, these spaces. And you talked about having a DIY approach to the spatialization as well. Can you tell us a bit more about that? Yeah, so the piece, latent spaces, I'm currently, well, for the past 18 months I've been a, um, a composer on a sound and music scheme called In Motion. And on that I've been researching how to, yes, set up my own spatial system at my studio. Because I think often spatial audio is, it's kind of kept in academia a lot of the time. Um, and I like to kind of democratize, um, technologies and kind of hopefully bring people into them. So yeah, I wanted to find out if there was a way to, um, set up your own sort of spatial audio system that's kind of quite cheap, but quality is, that is accessible as well. So we've kind of been doing that. So we've got A-A-D-I-Y 6.1 set up at my studio. Um, so I've got a sub for the first time carro, which is a Oh yeah. Doesn't that add an extra wonderful dimension to your life? I dunno how I lived without a sub before, to be fair. So, yeah, so I've got the 6.1, and I'm using Ableton actually to kind of spatialize this. I wanted to use, uh, yet a door that. Is kind of that everyone uses. So I'll be, I'll be sharing some of my processes through sound and music of how I've composed the piece. Um, and yeah, just the idea of thinking for me. I know you've worked with spatial audio as well, Caro. Mm-hmm. It is a different headspace, isn't it, to sort of think about sound being immersed within. Mm-hmm. Um, a setup. So I'm kind of thinking of it as in terms of like stepping inside the model, thinking about different states that you might encounter and how I might kind of play with clusters of sounds. Um, how you might feel immersed in a, inside a computer, for example. Um, the kind of placement of sounds as well and thinking about con sort of movement and panning. Um, so yeah, this is all happening at my sort of. Studio in Manchester at the moment, and then this is gonna be translated into the kind of more kind of textile heritage space that the exhibition will be in. And you are using spat, aren't you? The Yeah, so I'm using the spat devices. Yeah. So I'm quite new to those. Um, and I'm just gonna use those actually for panning certain sounds around. Um, I think what I've been learning is that. The more you kind of move sound around, the less effect it has. So I think just using some of those movements at particular points within the composition is sort of more effective. Less is more. Less is more. Yeah. And also having, another thing I was learning about is just having versions of things. So I think usually when you sort of, yeah, you composing for stereo, um, you know, you might have a couple of types of sounds that you might layer or kind of, you know, put around the. Stereo field, but I've been doing things like where I, where I'm recording a particular sound. I'll do lots of iterations of that recording. So I've got the ability to kind of bring in different, um, slight versions of it, whether that's through filtering, uh, or slightly different effects. So you're getting that, um. Kind of adding that spatial dimension to, to one sound across physical space. Yeah. So it's making me think about composition in a completely different way actually. But yeah, I'll, I'll be sharing some of those processes through, through the Sound and Music project. Wonderful. So did you find you had to optimize or upgrade your computer system in order to manage something like Spat and the Spatialization? And how did you, I think there's quite a few ways that we've done it. How everyone does it. How did you get round the. Able to master his only stereo at this point. Yeah. Challenge. Yeah. So I've been working with Guillem Doja, um, who is an Electroacoustic composer. Oh yeah. Brilliant. Um, and he's a brilliant audio techer. He's been helping me set up kind of like a template sketch, uh, in Ableton. So what we've got, 'cause it's a 6.1 set up, we've basically used the send and return section in Ableton. So that's what I did. Yeah. Did you? Yeah, and it's, it's been quite easy to do actually. So we've kind of set up all these sends. It's all about the routing, isn't it? So I can send, um, mono sounds to discreet channels, so one to six, or I can send a sound, um, to like one and two, and then pan it to one side so it becomes just sent out on two. Yeah, for example. So yeah, initially I was thinking about setting it up in terms of with six, having a front, so one and two having a side. Three and four and then a back five and six. But actually when we thought about that in terms of an installation, you haven't really got a sweet spot. So when you're producing that, that's fine. 'cause you've got that sweet spot, you're in the middle of this six, but in the installation, people will be, will be moving around more in the space. Mm-hmm. So I'm actually thinking of it more in terms of clusters of sound. So having sounds happening over in this corner, for example. And then things happen moving across to the, to the left. Corner, for example. So yeah, so setting it up in Ableton. Yeah, it's all been about the sends. And then on certain sounds, I'll have the spat plugin device, um, which will do kind of panning around the whole, the whole space. So there's kind of a unison feel when you're sending it out to all the channels. Um, but then I've used LFOs to kind of, um, yeah, to maybe like split the audio or have certain, um, yeah, certain swells happening at certain points. Um, make things modulate across, across speakers. So yeah, it's kind of send in returns, LFOs, um, and discreet kind of positioning of sounds is how I'm working with it. Yeah, that sounds wonderful. And especially that enveloping sort of, um, aspect you can get with that dance around the more immersive fields. Let's unpack some of your sculpture work. 'cause I see those as almost your, your instruments. Yeah. Yeah. And these also. Though you've come up with symbols, there's a whole conceptual thing around that. So yeah. Could you tell us a bit about that? Yeah. So yeah, I've been working with sort of sculpture for a while. Um, it's that kind of hands-on material that I really like. Um, so yeah, I think in the early days I would be working with sort of drum machines and Ableton. Then I'd have these metal sculptures that I would kind of use contact mics and mic up. Um, and they had a particular kind of, um, resonant frequency that I would sort of. Work within and kind of blend into the, um, the music that I was performing with. And I just like the sound of metal car carro. There's something about it. You're a metaler. I'm, I think, and I've, I've sort of thought about this, where this came from and I'm, I'm wondering if it's because I used to play trumpet, so I always had a piece of metal in my hand as a child. I don't know. Mm-hmm. So, yeah. So I've been working with, with kind of, yeah. Material sound. So, um, for a while. This piece, the kind of visuals, um, they are also made with um, an early visual, um, machine learning model called Style GaN. So I had a data set of, uh, electronic circuit symbols and alchemical symbols. Um, so this was kind of a visual data set and I trained with those materials and then I. Kind of the output of that was these kind of quite intriguing, um, symbols. So I've been working with these as kind of like a system, like a technical language for a while, and they're completely abstract, you know, but I guess for me they sort of represent this idea of emergent technologies and us as humans trying to find meaning and a language to work with these systems. So they sort of suggest certain things to me. Yeah. So in the piece, um, I have this process of making these. Symbols become real. So there are actual physical steel sculptures. They're about kind of 50 centimeters wide, um, about a meter tall. Um, and they resonate, so they're hollow inside. So I can send audio out to those through a transducer and kind of resonate those sculptures. So that's what I'm hoping to do for this installation. Have a point in the piece where there's kind of these, uh, swells and these kind of, um, yeah, these kind of low frequencies are sent to these. These sculptures and they kind of rattle, so you should be able to hear those. Um, yeah, and I've used them to sort of think about, um, kind of patterns and. And how, how symbols are just kind of another form of language. So I've kind of used them in terms of kind of programming and making kind of like rhythmic patterns with those as well. Um, and through tone. So yeah, so they're working in a few different ways in the piece. They're part of the composition, but it's like the visual language for the piece essentially. Cool. Yeah, it's really interesting listening to you. 'cause I suppose I had my own bias of like, oh, AI is this. And um, you know, in the environmental aspect, the ethical aspects that we've kind of touched on, but it actually is reminding me of what people said when computer music first came in and everyone going, they're gonna take our jobs. And I mean, obviously there's other angles on this. Um, you know, I don't wanna be simplistic or naive, but in terms of that, it's sounding like. What you are saying is, yes, it can be evil or good, whatever that kind of, you know, reductionist sci-fi sort of approach. But actually it's how you want to build that relationship with that technology. So, for example, I try and buy secondhand stuff. Um, I try not to buy too much kit and there's all the plastic, there's all the minerals we are using, there's all the batteries we're using. There's the cost, you know, the environmental cost of what we're doing with technology. Anyway, so I think it's interesting that hearing more from you, how. It's like anything else, it's, it's the way you want to find in. Does that make sense? Yes. It, it absolutely does. Carrie. Yeah, definitely. Um, I think that's it. I think we're just, we're, we're currently in this sort of latent space era. We've always had, you know, if you work with electronic music, there's always been an element of automation. We've worked with algorithms for. For years, you know, this is just a new form of emergent technology. And yeah, just through kind of, yeah, media history and new inventions, there's, there is always these kind of hopes and fears we we're sort of keeping up with these technological advances. I think so I think it's for us as humans to sort of make sure we legislate properly, we protect people, we look after the environment and sort of we're comfortable with our own ethical use of these technologies. Um, so for me that's. That's really important. Like I say, I, um, I only work, I only use these systems with my own materials and that's kind of what I'm interested in, that kind of human machine collaboration. Mm-hmm. Um, I'm interested in the errors, um, and how things can go wrong with those as well. Yeah. So I think you can work in a more ethical way with these systems and like, yes, like we say, the wider context there is all this AI doom, but there is also AI hope as well, you know, in terms of the medical. Um, system. Um, there's lots of advances in these technologies as well. We just need to place, we need to kind of keep in check who has control and who has the power of these systems and find alternative ways, um, to use these creatively. And as an artist, that's a wonderful way to live out your politics as well, isn't it? Yeah, I think I've always been concerned with, um, yeah, being critical of, of technologies and how we use them. It's about a relationship and what you feel comfortable with. And like I say. These materials that I'm using, they're kind of, they're already like four years old. Um, so I haven't done any new training for this, for this piece, for example. Um, and it's interesting to think that that is now these, these particular models are kind of already of the past or already of that history. They're kind of retro in a way. They sound, they sound quite old. So yeah, it's interesting that we're already kind of moving past some of these early models as well. Yeah, and I think that's kind of the history of technology, isn't it? That uh, yeah, things emerge. Then new forms replace that as well. And we kind of have to find a way to not just live alongside, but um, keep in check and use these things. Yeah. How we feel comfortably and, and respectfully. Yeah. Wow. Definitely. Yes. Food with thought. I wonder if there's any kind of emerging next steps for you or next following that fascination or that curiosity. Where does it feel like it's leading you next? Yeah, I mean, I think I've, I've been thinking a lot about the sort of physical materiality of, um, yeah, how I work with machines and materials a lot, and it kind of comes down to systems. I think I've been thinking a lot in terms of, so when I start a project, for example, I always kind of draw a block diagram of how I think, um, a performance system might work. So yeah. Where's the computer? Where's the sculpture? What am I doing? And I've been thinking a lot more around actually the sort of geology of our technologies, um, and how these have like a material memory and a history. So yeah, I'm interested in exploring more the systems at play within the actual materials of our technologies and seeing how I could maybe bring that more into. Kind of what I'm working with in terms of composition as well. So yeah, I think thinking around those sorts of themes around the sort of geology of our media is where I'm sort of looking to next. Cool. And does that involve neural synthesis? It could potentially, yeah. Right. It could. But not necessarily. Not necessarily, no. Um, if that feels, yeah, if that feels like a, a, a sort of tool that I want, want to use. For example, maybe looking at ways that you can be how lo-fi you can get with machine learning. So could I have it on like a raspberry pie in my house, for example? Yeah. So thinking more in that sort of low-fi. System way. I think that approach. Awesome. Wow. Well, I wish you all the best with your continued adventures in, um, technical and creative wonderment. Thanks, Carrie. It's great to speak to you. Thanks for having me. Thank you for listening and be sure to check out the show notes. For further information as well as links and details of other episodes in the electronic music series. And just before you go, let me point you to sound on sound.com/podcast so you can check out what's on our other channels. This has been a Caro C production for Sound On Sound.