Uses of Algorithms & More w/ Woulg, Super Producer & Educator
Written by: Emily Cinder
Greg Debicki, known artistically as Woulg, is a Montréal-based creator who blurs the lines between sound and visual art through his groundbreaking work in glitch music and experimental sound synthesis. With a strong New Media and Music Composition foundation, Greg explores the depths of sound manipulation and custom-built audio tools to challenge conventional artistic norms. In this interview, Greg dives into his unique creative process, the inspirations behind his experimental techniques, and his vision for the future of digital art and music.
Synapse: So I'm writing for "The Synapse" which is a zine and online reading platform for neuro.studio … You made a post about how you wanted to do a class for production in conjunction with ChatGPT and Python, and it prompted me to have this conversation. We could start from the beginning here. I read that you're from Canada, but you've also studied in the United Kingdom. Do you want to share more about what that experience was like? I also know you for your portfolio as well, and I love that you just have that whole resume out there. It's pretty impressive, and I think it’s cool that you've taught at colleges and in a variety of settings.
Woulg: Okay, so yeah the school that I went to, right! So I did my undergrad in digital new media art and I was doing installation sort of pieces in my work at that point. And then when I finished school, all this stuff that I was really interested in, I was having a hard time kind of visualizing how to keep moving that forward. I did quite a bit of work with getting different arts grants to do more installation work, but it just felt like this sort of, in some way, unsustainable art practice and I still couldn't really imagine how to sell that kind of stuff to like a gallery, or show, you know? And since I was always interested in music, I ended up kind of pursuing that. That school that I went to in England was actually this historic art school where I went to study for a year on an exchange program. It was a really funny coincidence that I ended up there. It was a last minute thing when I applied to go to school there and I had remembered that somebody had came from that school to my school on exchange, and he said that their school was on the beach, and I was like, ‘Oh, that would be amazing to just go study and be by the beach for a whole year.’ But what actually happened was, that particular year, that school that I applied to had acquired this historical art school, the Dartington College of Arts … so they had just like, kind of amalgamated that school into them, and that school was just like in the middle of the countryside in the middle of nowhere in Devon ... In the end, I think I learned more in a year there than in art school up to that point in a way, because it was really a ‘go at your own pace’ sort of thing. They were really just like, ‘Go make art,’ and then you would go and make some and then they would come back and just give you more information. But it was really, really cool. It really pushed me to kind of get out of my comfort zone and try different things. I was working with dancers they had, for example … because I was in the music stream, I was supposed to have some one-on-one lesson time that was allocated to me for practicing my instrument – but as my instrument, I chose the computer. They arranged for a professor from the University of Exeter to visit me weekly for one-on-one lessons, exploring the concept of thinking of your computer as an instrument. He's such an awesome guy. He was just such a fascinating and wonderful teacher. So, he taught me about all these different ambient artists and about all these kinds of computer music, through this vein of, like, installation performance art but using the computer as the enabler of all that stuff. I think that a lot of the way that I think about music is through the lens of new media art … I still don't really know that much, like, music theory stuff.
Synapse: I hear a lot of people say that. People who’ve been studying this kind of stuff for so long, and that they're kind of just like moving through it intuitively and building off of the things that they've learned from other people, and then with that, some information gets lost because you don't know everything, but then you're teaching someone else something and it's like they're just building off of that, and then teaching someone else the next thing. It's sort of like whispering history in someone's ear, and a little piece of it reaches the next person and that can transform into new techniques or concepts. But that just goes to show you, maybe there's not as much relevance in conventional techniques being set in permanence. There are obvious foundations but maybe we should question what is considered to be good or professional, like, these are all just things we come up with in our mind's eye, you know?
Woulg: Yeah there's this sort of, like, illusion of the expert, right? Actually this – speaking of, kind of touching into the AI stuff – this is one of the things that I think about a lot with using ChatGPT to learn stuff, is that people are like, ‘Well, ChatGPT hallucinates and stuff.’ It’s like, yeah, but your teachers in the university also hallucinate. Like, they get stuff wrong also. Or they hold on to some belief and they pass that belief on to you. Then later down the road, their belief gets updated, but it's not like they email you like, ‘Oh, well just FYI. 10 years later, turns out I was wrong about this thing,’ you know? It's like you have to discover that on your own. So to me, I still find ChatGPT is such a great learning tool. Because, you know, of course you gotta check [what it's saying]. But the same thing goes for any teacher, right? If you take lessons, or whatever you learn from any teacher in university, you can be sure that they're getting most of it correct, but you shouldn't believe that everything they say is 100% correct all the time. You know, you still have to use your logical reasoning.
Synapse: Yeah, definitely. Even under GPT's prompt container, it'll be like, ‘Hey, ChatGPT is not always correct.’ And like, what do you think? What's your opinion on there being an art form to prompting? What have you found with that, because I was hearing about it in that context and I obviously am using the tool. I've been using it for a while now, and then I've noticed that I've just gotten really good at prompting and getting those results. So what's your experience with that?
Woulg: So I did this project for most of last year where I kind of decided to spend, I think, around 20 hours or more a week trying to figure out how to just get better at understanding what is going on with neural networks and try to make my own. I set myself this project of making a variational autoencoder that would control where you would learn the parameter space of the operator in Ableton, and then you would be able to move around in the latent space like this kind of like 2d map, and not just access any settings of the device, but also smoothly interpolate between them. I never ended up succeeding in that. I had a couple pretty good promising results, but I never got to the result that I was kind of hoping to get. But it taught me a lot about prompting ChatGPT, because I was using it to learn how to write, you know, and how to do a lot of this stuff. I also had some help from friends who know more about that stuff. The biggest thing that I found is it's really important to make sure and to really consider what assumptions ChatGPT might be making (or whatever language model you're using) might be making about what you're trying to do. A lot of what we say when we're talking, when we're sort of writing in this way, there's a lot of hidden assumptions that you make that you might not realize that you're assuming. We just sort of hope that, if you're talking to a person for example, you just sort of hope that their assumptions are based on the same thing. But sometimes those assumptions are things that would be obvious … it doesn't have that context, but it's gonna pretend that it does, right? Because that's how this predictive [stuff] works, right?
Synapse: I didn't think about it like that.
Woulg: In order for it to generate more text, it has to work on some assumptions, right? So if it doesn't know what the context is, it just kind of guesses what the context is based on what your prompt is. So if you're trying to do something specific, it's really important to think about, ‘Okay, what assumptions am I making, and making sure that I'm making that clear? What context might it need in order to do the thing that I actually want it to do?’ I think it's gotten a lot better at this, but before it was very forgetful. So you would get a couple chats, you know, you'd get a couple responses down and it would have forgotten what happened before. It’s not like it was going to tell you because it doesn't know that it forgot something. It just is continuing to generate, right? Before, I would often have to keep reminding it, ‘And remember that we decided this before, and remember that the code looks like this, and remember …’ you know what I mean? It's like working with a really, really knowledgeable teacher, right? Like this has access to this huge knowledge base but they're really forgetful. They have these ‘personality quirks,’ right? Not to anthropomorphize it too much, but I think for me, it's helpful to sort of anthropomorphize it because it helps me think about, ‘How would I deal? How would I interact with a person with these qualities?’ Which, I feel is easier for me to think of ideas of how I would interact with a person in that state, than thinking too much about it being an AI or whatever. Then also the other thing, I guess like the last thing is, thinking about what it is likely to know or not know based on the data that it had access to, right? So one thing I very rarely will ask it for is stuff relating to music, right? I don't like to, I find it almost always gets stuff wrong. If I was like, ‘Oh, I would like to get this sort of effect. What kind of effect chain would you recommend?’ And my reasoning for why I think it gets that sort of stuff wrong, with digital audio stuff, is we're in this place where a lot of people who use digital audio tools (such as using DAWs and plugins etc.) are not coming at them from the perspective of understanding the underlying mechanisms, right? It's not like you need to understand the equations for a Butterworth filter in order to do high pass stuff. Like, you don't need to understand the underlying tech in order to use it. Most musicians don't, and for good reasons. You don't need to understand all that stuff. But unfortunately, the position that it puts us in is that there's this sort of disconnect between what the tool is actually doing and how we talk about it. So there's a lot of times where, if you look on Reddit/internet for an answer, there's a lot of stuff that's confidently wrong.
There's tons of threads of people just being confidently wrong about stuff and ChatGPT was trained on a lot of that sort of stuff. Whereas, like with code, when you have the underlying thing that they're talking about, is a text thing – that's included in the statement or question and ChatGPT has access to that. When we're talking about audio processing stuff, the underlying technicality of that audio question we’re talking about isn't there. So the language model learning doesn't have access to the real thing that we're doing, right? It just has our conversations about it … yeah, the last thing about how audio tools are made. A lot of the times these tools are made in this way where the underlying processes are unknown or obscured.
Synapse: Yeah, like a nice UI – it wants to make you feel like you're having an experience, rather than understanding what's going on, right.
Woulg: Right, exactly, right. So, yeah I do the same thing too, right? With my Max for Live devices, I'll often name a knob like ‘squishiness’ or something, and it has little to do with what's actually going on, or it's controlling multiple parameters. So this again makes the discussion even more confusing, right? Like, if you have a whole discussion about people online being like, ‘Oh, you have to turn the squishiness up to up to 11,’ and then other people are like, ‘No, that compromises the crunchiness,’ or whatever. We're talking about abstractions on abstractions on abstractions and everything's this slippery kind of mirror world, right? So to me, it's like no wonder ChatGPT gives you shitty advice for audio processing stuff. The thing that's cool about it is that it does know the underlying processes, right? So it knows the technical stuff, because it's also trained on that. So if you wanted to know, for example: Say, you're really confused about compressors. which I feel like in this day and age was more of a meme from back then, but let's say we're still confused about compressors. So you asked ChatGPT, ‘How does a compressor work? Or like, what does a compressor do? Or, what are the different knobs on a compressor?’ I think asking in that way, it’s likely to get it right. Especially if you phrase it in a way where you're interested in the underlying process … ‘Okay, but how does it do that?’ Then if you kind of get it in this zone of talking about the stuff that it actually knows about, which is all these technical aspects of how compressors work, and by understanding a bit more about that, you can understand ‘how do I use it in my music,’ right?
Synapse: Yeah, I mean, that's going back into kind of you talking about how it's not really good with asking it music questions, and because these conversations get so diluted. When people are starting out with chat, or they're not really used to using it all the time, I feel like they are asking things that they would ask from a personal mindset compared to maybe someone that already knows what they're referring to. ChatGPT just can't handle that, because it's a chat bot, essentially – it's like neural networks. That's really cool that you were trying to make something like that. Were you just entering code into an IDE? I want to experiment … I've been attempting and have started a million projects with little to no results. I do want to try to make something like that sometime … I guess, like, Python is really good for doing stuff like that.
Woulg: Python is great – in terms of the AI stuff in general, a lot of the tooling is built-in in Python. So because it's the main language used by researchers, right? (As far as I know). In terms of just bare AI stuff, Python is kind of where it's at now. But if you want to get into the audio stuff, Python's not really meant for real time audio or anything like that. It does have some tools and some libraries that you can download to do that kind of stuff. But it's not really its sort of main “language” for that.
Synapse: Object oriented, right?
Woulg: That's a different thing.
Synapse: I was reading that C ++ is used a lot for, like, making plugins/VST's, like the back end of making plugins. So I guess that's kind of what brought that question about.
Woulg: So yeah, C++ is often used for making VST's. Python is an interpreted language. C++ is a compiled language, it's also a lower level language so you can be more optimized with it. And when you're doing real time audio stuff, your code needs to be quite optimized. Not saying that you can't do any real time audio stuff with Python, just that, there are better ways to go about it. I think there are easier ways to go about it than doing it in Python. The reason that I put out about that workshop of using Python to get into this stuff, is that first of all, ChatGPT knows Python really well, in my opinion. There's people who are really good at Python, who would disagree with me there, right? But I'm not talking about writing the most amazing optimized code, or whatever. I'm talking about getting into programming.
Synapse: What does it mean to you for code to be optimized? To me when I think about optimized code, it refers to code that someone else is also able to understand as well as the author/programmer. But when you're just writing your own code, it's just a mess. It's just messy and unorganized. I mean, you could try to be really good at being organized, but I think people get so deep in their projects, it can get out of control.
Woulg: Oh yeah, yeah I know you mean. There's like, optimizing. I just sort of mean like, I don't think that ChatGPT is necessarily going to give you the thing that runs the fastest, and also sometimes isn't necessarily going to give you the most readable code, right? Like, from that perspective either. But the thing that I find that it's good enough at generating Python, and as far as I know, that's the thing that it's kind of best at. That's the language that it's best at. And so it's a great starting point for producers who would want to make some custom tooling for themselves or for their music making practice. I think Python is a great place to start because there's a bunch of tasks that are related to making music that you can totally customize. So for example, let's say I wanted something to organize my sample library in a different way. One thing that I'm really into are these dimensionality reduction techniques. So for example, if you analyzed all of your samples for lets say, pitch detection, how noisy or tonal they are, whatever texture, and then you use that and then you do like a dimensionality reduction of all those kinds of dimensions on your samples. Then you could plot all your samples in 2D or 3D or whatever, and then you have this kind of correlation of how similar your samples are to each other. Not only is it fun just to get into because you're learning all this cool stuff about dimensionality reduction and dimensions and stuff which is fun and interesting, right? But you're also giving yourself something that can be a super useful tool for when you're making music, and you can also do all sorts of offline audio processing tasks … for example actually, one common example that I think is getting more popular, is concatenative synthesis, right? Are you familiar with this term?
Synapse: I don't think I am. No, like I feel like I've heard of it from a friend, but I don't know exactly what it is.
Woulg: Yeah, it's kind of a new exciting area of research for audio synthesis stuff because of this paper that came out a couple or a handful of years ago now. So you have a target sound and you have a source sound. The idea is that you want to reconstruct the target sound with the timbres and the vibes of the source sound, right? So the classic example of this is you have your target sound of "Let it Be" by The Beatles, and then you have your source sound of a bunch of bees buzzing around. And when you do this concatenative synthesis, you then have this output sound, which sounds like a bunch of bees buzzing to the tune of, "Let it Be" by The Beatles. Which is really trippy and cool, and you can imagine all kinds of interesting uses for that, right? So say have a break beat, but you wanted the break beat to sound like a bunch of broken glass, so then your target is your breakbeat, your source is your broken glass sounds, and then you use this concatenative synthesis process to generate a breakbeat out of breaking glass sounds.
Synapse: Is this something that's used in newer samplers? I've experimented with some newer ones and it's like, you're taking a bunch of samples, and then kind of mixing it together, and then you can customize the parameters of how it's mixing itself together. But maybe this isn't the same thing, but it just reminded me of that.
Woulg: Yeah. So the idea of morphing sounds together … this is great. In particular this is a really great example of the thing that I was talking about before, where, like, the real underlying process is hidden, under this term of morphing, where we all feel like we know what it means to morph between two samples, but the technical underpinnings of that is like, super ambiguous. I don't think there's a set definition of what it means, technically, to morph between two samples. We just have this sort of intuitive understanding of what it would sound like. And there's a bunch of different approaches to getting that to work. A classic one is to interpolate between the two spectrums of the sounds and there's a handful of different ways of doing that. And there's many different ways of interpolating between two or more samples. The concatenative synthesis thing is interesting in that it's this sort of classic approach where you do this granular thing where you split the target and the corpus into like small grains. Then you analyze the grains of both and you just do this like patch replacing. So you replace all the samples – all the grains in the target – with the closest match in the source. The kind of research that came out I previously talked about, it uses this different approach of like taking parts of the spectrum and like using this technique called non-negative matrix factorization, where it takes pieces of the spectrum of the source and kind of places them in a nice place, like replaces the parts of the target that way. It's a more complicated thing, harder to sort of explain in a one line, or anything, but same general sort of idea. If you want to hear a really good example of that, there's the album "Zero Point" by Rob Clouth, which uses this new concatenative method. I did an interview with him on one of my channels.
Synapse: Yeah, that's super interesting. I will definitely look into that album and look into that interview. Yeah, that's so weird. I mean, there’s unlimited possibilities and that feels like artificial intelligence in itself. That feels like something that is related to some Mandela Effect type of thing. Something that I wonder about you is how do you go about finding research that interests you? I'm interested, because, like, I'm a person who is always seeking knowledge, whether it's taking a lesson, or just doing my own research, or being in school on and off for a long time, whenever it's accessible for me. And I'm sure there's other people in the same knowledge-seeking sense and path, you know. How do you go about it? Like, what goes off in your head where you're just like, ‘I need to figure this out.’
Woulg: I had this sort of revelation actually, it's funny that you asked that. I had this sort of revelation in the last year where I was sort of like, ‘I'm not actually sure if my main curiosity with music was to learn how to make music,’ or whether I just really wanted to understand how does all this work? Like, what's going on underneath the hood? Like that you're able to change sounds in this really cool way, or in all these crazy ways I guess. I'm actually not sure if my main motivation all these years has been to make the best music, or if it has been to understand what is going on with digital audio processing, right? But to answer your question kind of more directly, I think, I have always been quite curious, and I like knowing stuff. I also find that when you're really curious about something, and you get that kind of tunnel vision where you're trying to understand this thing, it's really the best distraction from the suffering of existence, I guess. You know, it's flipping through memes or watching reels or tiktoks or whatever or even longer form YouTube videos, and it's like, that's distracting for a bit. But if you really want to get distracted from the suffering of existing, I feel like really trying to understand something really confusing is, it just hits the spot much better, you know?
Synapse: Yeah, I feel like people that just deny that are self sabotaging their excellence or something, which we can all go through in life. That's kind of another thing I want to ask you about, is just like obstacles. Obstacles you've come across with just, you know, doing what you do. And obviously everyone runs into obstacles, I'm sure, but yeah how do you keep pursuing this? How do you keep pursuing knowledge and figuring out what makes you tick?
Woulg: I think for sure there's been a lot of obstacles. I think it's important also to acknowledge that I've had a lot of privilege, right? Like my parents for sure they weren't rich, but we were okay and also they were really encouraging about me going into the arts. If I was curious about something, they would support whatever thing that was.
Synapse: They are constructive and supportive parents.
Woulg: Yeah. So that already is a big leg up. In terms of struggles and difficulties, I think the main stuff has always been that I've always had troubles with my mental health and also having more time for music. Like, money has always been a difficult thing, because I was always sort of trying to get away with, like, what's the minimum amount that I can earn in order to have the maximum amount of time to spend doing all this stuff? And when you build your life that way it can be complicated, right? those two things together … there could be definitely lots of difficulties. I would say that the mental health stuff is probably the biggest hurdle, without getting too deep into it, I think it's always been something that I've struggled with. I think for me, a big part of learning how to make music was learning how to deal with my brain, right? Like, deal with the way that my brain works and trying to figure it out. Okay, so today I don't feel motivated at all. How do I deal with that? Is it best to take a break today and then get back to it tomorrow, or is it best to push through it today? And how do I motivate myself to push through it? These kinds of questions of like trying to figure out how to work with your brain to get the most juice out of it, I think are the biggest lessons that I've learned from making music all this time. But I also feel like I just burnt myself out super hard. At the beginning of last year, I stopped doing shows, and I was just like, ‘Okay, I need to just focus on what am I gonna do to become less anxious.’ And also trying to make a little bit of money to make my life more comfortable and stuff, right? So, I wouldn't recommend doing exactly what I did. But then again, I don't know the struggles of doing it any other way, right? So it's not like I can advise on that. I have some students who, for example, their priority was to have stable income and then put the music kind of on the side, right? So first thing was to get a steady job where they're making a comfortable amount of money and then slowly building up their music career. And I can totally see the advantages of that, right? There's, a lot of things where just having the money to sort of invest in yourself, invest in your tooling, invest in your situation, helps a shitload, right? But with the downside of that is you have less time to practice. So it's the balancing act, right?
Synapse: Yeah, I'm feeling like I'm going through that right now. I just want to further my knowledge, brain, and education. I think that's personally my fear, I think is running out of time, or just like not being able to further that knowledge because I'm so busy trying to be stable in life and just maybe I'll die tomorrow and just not know anything. Like, ‘Damnit,’ you know what I mean?
Woulg: I don't know if that ever goes away.
Synapse: Okay, that makes me feel better. I don't want to come from a place of woe is me in terms of thinking I'm the only person going through this. It makes me and probably many other humans feel better to even know that someone of such an accomplished and intellectual caliber, or just someone that's just studied something for a long time has this same dilemma with time and their work.
Woulg: It's comforting to me to hear that you're also going through that, because, you know, it's definitely not woe is me. There's a difference between knowing that you're not the only one going through it, and like, you know, somebody being like, ‘I'm going through it too.’ I mean, there's like, a tangible difference, I think, between those things. So even just you saying that makes me feel a little bit better as well, because that's something that I always struggle with too especially now that I'm getting more into AI. Like I just feel like I'm constantly behind. I mean, that's the whole AI world right now, everything is getting better, so much faster, and there's so much money and resources and super smart people in the field that it just feels like every day there's some revolutionary new thing that somebody discovers, or whatever.
Synapse: Maybe because they're using AI to help them?
Woulg: Yeah.
Synapse: I've been using it a lot to streamline a lot of workflow things. I'm one of those people that red lines, um, and spreads themselves thin doing a million things. Last night I was thinking of how I couldn't wait to feel as though, ‘I'm totally available and free’ but then I'm like, is that really what I want? Like, would I really be who I am if I wasn't constantly starting new projects? And, I mean sure, there's things in my life right now that I'm sacrificing, things I don't want to do, you know? I feel guilty for using it sometimes, because it is improving my life so much, and it's helping me get things done faster. I've been using it to adventure into the CSS and JavaScript side quests. I am kind of mostly a front end person, and I have been discovering all kinds of insane visual type renders. People are just building 3D models with JavaScript, and this is like, blowing my mind. But just stuff like that, I'm like, ‘What am I gonna do? Like, go online and ask Google and ask them for help with this type of code, or am I just gonna generate probable structures regarding these libraries and just agitate the code to my own style? I'm gonna do that.’
Woulg: It sounds to me like that you have this sort of, like, kind of hacker, anarchist, sort of mindset, which is exactly how I feel too … I mean, that's part of the thing that I think has been really liberating about being able to use ChatGPT. Okay, I'll give you a really simple but tangible example. So for my lessons, before, the way that it worked: someone would email me, we'd pick a time together, and then I would schedule the Zoom event, and then send that invite to them. Then when the lesson was over, then I had to look at my calendar, type in all the invoices and send them off. Now through ChatGPT … I just have this one Python app that I made using tkinter for the GUI or the UI and it just shows all the lessons that I have for the week. The ones that are today are highlighted. When I have a lesson coming up, I press on the button, it opens up Zoom, it goes to that lesson. And when it's done the invoices all get put into a thing. I run a separate script at the end to send all my invoices. So it's totally automated it for me. Before, every week I would grind to just like, get that done. And so now that I have it just as a script, I save so much time just with that, right? Like, it's just like an hour, at least an hour every week that I’ve saved minimum, right? And so these sorts of, like, little tooling things – it kind of reminds me a little bit of this thought that I had, of I was looking for desks for my studio, and looking through all the desks, thinking about them, I realized that I think my ideal desk setup would be one that I custom designed and built exactly for my space. Everything would have this custom tooling. But the cost of trying to do that, is either you hire somebody to build that for you, which is totally outside of my price range, right? Or the opposite is like you sit down and learn how to do it, which is more interesting to me, but would be such a huge time investment because I'm not like a woodworker, carpenter person, like, I'm terrible at all of that stuff, right? Yeah, it would be an ugly piece of shit.
Synapse: Yeah, I feel like I go through this a lot too, honestly. I really believe in this customization of everything in life. Like, there's this guy I saw on this video and like, he puts red tape on everything or the color red in relation to objects around him. Everything in his freaking house is red, even red duct tape to label his cabinets and I'm like, ‘This is some over compulsive shit that I'm so into.’ I go through that with, like, even clothes and studio items. So I guess I'll try to figure how to sew and 3D print hardware enclosures now I guess.
Woulg: My god, yeah, the amount of times I've bought a piece of clothing with the intention of I'm gonna modify it, totally destroy it, and then just totally screw it up. Like, sometimes, occasionally for the better, but usually I just mess up that piece of clothing … This is something that I feel like I have to remind myself pretty much every morning where I'm like, ‘Okay, what's the actual thing that I want to be spending my time on?’ Because otherwise I'll get distracted with customizing everything, right? I was doing some freelance work where I was designing this like, visualizer thing for audio, and because I had to make it work on Linux, I got Linux subsystem for Windows, and then I started getting more into the terminal. And I started customizing my terminal, and like how my stuff worked and then; and I just got so lost in this thing of bash scripting and learning how to use Vim motions and then I had to really pull myself away from it because it’s so tempting to just overdo and customize it like that. Just to make it feel right.
Synapse: Right, it needs to feel like your tool so you can use it efficiently and effectively. And I totally understand that there's gotta be some kind of balance. I think.
Woulg: Finding that balance is hard. I'm really glad that we got to have that conversation. Makes me super hyped and I imagine that there's a lot of people that are reading that feel similar.
Synapse: I'm really happy that you're down, we have covered a lot of cool topics already. I was gonna ask, what would your advice be, as a programmer and someone that also does audio and visual synthesis, if someone with limited background or funds, but has goals to blend those worlds together, what’s the initial move for starting this? Because a lot of people I know use Ableton or FL Studio and for example they use a Mac, and especially if they perform live, because they need something very reliable, but then there's a lot of people I know that are just programmers that use Linux because of its customization. There are many ways to go about setups and sometimes people have to work with what is available to them. So, where should someone start? Especially, let's just assume that someone doesn't really have a large budget, how can something like this be accessible to everyone. I'm really pro-education and making resources attainable despite where one comes from and you brought up the whole anarchy aspect of me. So with that said, I am curious what you think about accessibility with these specific skill sets.
Woulg: Yeah, amazing question. So right now I'm using this Windows laptop, and it's okay, like it's like a gaming laptop or whatever, and it's okay. I do find myself wishing that I had splurged for a Mac, but you know, at the end of the day, you use whatever you have the budget for. My way of writing music has always been just the computer keyboard and the mouse. I never have had a MIDI controller plugged in when I'm writing music. It's all just like clicking and moving stuff around in the DAW and in Ableton. I used FL Studio for a long time and I really liked it but then when I started teaching, I switched over to Ableton because I found it easier to get students who were interested in Ableton, and then I just kind of didn't look back. I just switched to Ableton basically, and then I use Max. I'm scared to rely on Max for Live (Max4L) and I think it's maybe just because I'm older and have seen the struggle of the integration of Max4L.
Synapse: Right, I wouldn’t know. I've barely experimented. I try, and then I'm like, I give up because this just doesn't feel like a fluid workflow for me.
Woulg: Absolutely, I feel you on that. Also generally for my music I'll use mostly stock plugins. I find they do the stuff I need them to do and I like that they're just sort of built into Ableton and I don't have to have another window open or whatever. If I'm trying to really EQ something really nicely, then I'll use Pro Q by FabFilter, or if I'm doing some mastering stuff, I’ll often use Ozone or some of the FabFilter tools. But beyond that, there's not a ton of VST's that I actually use, just because I find that it kind of messes with my workflow. I don't know why. It's just the way that my brain works. I just like to have the tools that are built in, and I just screw with them.
And then when I play live … I didn't have a MIDI controller. It was just like my computer keyboard and my trackpad that I would use live on stage and that was my midi controller … It's tempting to buy a thing with like, a bunch of knobs and sliders and stuff and have a dedicated tool, because it's nice to have that tangible thing, right? But so often we sort of forget that our laptops have these human interface devices. A ton of money and research has been put into building these things, making them smooth and streamlined. So it's awesome to be able to use that to make music. Back in the day, Mac laptops had a tilt sensor in them, and it was so that if you had an actual hard drive in there (not an SSD), and you knocked the computer off the table, it would pull the heads off of the platters in the ArcGIS to prevent you from damaging it when the computer hit the ground, right? So it had this accelerometer in there, and you could hack into that and just get that accelerometer data. So what I used to do on stage was I would have my laptop, and then I would just be tilting it like to control the music at certain times. When I moved to Windows, I just built myself a little controller with Arduino stuff. You can buy, a $3 accelerometer and like a $5 - $10 chip. You put them on the same little kind of like perfboard, solder it together, and then you've got that same controller, but just as a little USB device. So I just had that, like, attached to my USB thing and then I could do the same thing. So yeah, and all of those things are low cost, right? Like, it doesn't cost anything to do this, to use your your keyboard and your trackpad as your interface because you don't have to buy these overpriced MIDI controllers.
Synapse: Yeah, accessible for sure. Yeah, yeah.
Woulg: So like if someone was curious about this, but I also would say, like if you're curious about this, I would recommend trying to learn how to build it, right? Because by building something like this, you're learning all these tools that you can then use to build other stuff. So instead of just getting a tool and using it, you're like building up your own skills to be able to make your own custom tooling and stuff, which I think is super important.
Synapse: Yeah, this could be, like a first project type of thing to kind of get into the realm.
Woulg: The other thing that is really good for that – video game controllers are fucking sick! They're expensive, but often people have them because they play video games, right? So that's another great music controller that you can use, that you're sort of like reusing. Instead of buying yourself a new MIDI controller. It's harder doing it this way, but when you're playing on stage and you have some custom thing that you built, there is a cool thing that comes along with that. It's a custom rig that you made, right? And then on top of that, you're learning tools, you're learning how to do skills that will make your life easier in the future. I just explained to you that keyboard controller thing. A great starting project: let's say you wanted to use Python to make a replacement for this, because you totally could, right? How do you make a Python script that takes any keyboard press and puts it on a grid? How do you make it so that it makes a slider? These are all interesting questions I think and it's not something that would be impossible … I think that would be a really great first project, actually.
Synapse: Cool! Yeah, that's so interesting and very accessible. I like how you touched on the Arduino. I'm trying to get more into microcontrollers and stuff, like integrating that with music. I'd love to interview more people that are interested in that kind of stuff. I am honestly unsure how large the community is when it comes to renegade or guerilla leftfield rigging / audio synthesis … Something I think about a lot with technology is, it's just a projection and reflection of the human mind … AI is just a newer thing that's going to make our lives a lot easier. And there have already been things that have come for years before that people have worked so hard on that make our lives easier that we don't even realize that we're using maybe.
Woulg: I think there's important ethical considerations for how this AI stuff is being used and developed, and also ecological concerns about how much electricity it takes to run these things. I think those are all valid concerns. Also, the idea of AI systems sort of displacing people's jobs, right? I empathize with that but I don't necessarily think that stopping AI is the appropriate response. I definitely wouldn't want to say I'm 100% pro-AI in general, I think there’s much more nuance to it. Trying to stop AI could maybe be better diverted into an adjacent field of like, changing how copyright law works, that would be an interesting side place to go … The last kind of thing to tack on there in terms of AI safety, when you have these big systems being developed, the thing that they're trying to optimize for is making money. Being first to market is really important and cutting corners – we've seen through all of late stage capitalism and probably early stage capitalism too – cutting corners is how you make more profit. Cutting corners on safety is one that has a huge history behind it. And so when you're talking about making these systems, I think there's potential for them to cause problems in the same way that we saw with Facebook recommendation algorithms. I'm trying not to get too distracted on that, but I think we've seen lots of history of this stuff going wrong. So I think that there's really valid concerns about AI safety, and how we structure the kind of scaffolding around that to protect ourselves from more shitty results. Either from bad actors using it or the profit motive optimizing the wrong thing and then, ‘Oops! We made something that’s really bad for us and it’s hard to undo. So definitely, I am empathetic about all that stuff, and I'm interested in it in my own little world.
Synapse: The whole thing with Meta right now and Instagram popping off with their art. I feel like I'm having conversations about what I'm doing with AI, which is completely different from what other companies, like full on corporations are doing with AI, and I think that's kind of also why I wanted to have this conversation, because I think a lot of the public might think that all of this is the same thing, but it's not the same thing. There are other avenues of utilizing this or discovering this. And that it's not the same thing that what someone else might be using it for.
Woulg: Because of all the money involved in these big models being developed, there's really good research into the model architectures and I think that the advantage of that is that there's this potential for us to make smaller, but a lot more specialized models that run on our own computers. So for example, if you have a model that generates music and it generates any music that you want, that's really cool, right? But if you wanted to do something really specific, then I don't know if you would fine tune that model, or if you would make a smaller model that's specific for the task you want to do. I mean, we saw that with ChatGPT, right? Just because I want to generate one specific thing, and ChatGPT wasn't trained specifically on it, that doesn't mean that it can't do a good job at generating that sort of thing. One thing that I am sort of afraid about is that, with all these artists trying to pull out their work from what would be the corpus, my concern is that you would already have small communities sort of disappear from this database, which I think it would benefit everybody if that data was in there. Now is that a personal sacrifice for the greater good of a company's product that they then could sell back to us? Yes, and that sucks. But for me, I feel like the concept of making money from my music directly, like that dream died for me 15 years ago. There's never in my brain the idea that I would sell the files that make my music, as if that would be the thing that makes me money. When did Napster come out? … I often think about this question of, ‘how do you want to interact with the world?’ Do you want to interact with the world is? Based on what actually is happening? Or do you want to interact with the world as you think it should be, right? So, for example, if you think it should be possible to make my living off of just selling my music – then you can fight for that. I think we need both sides. I'm more on the side of interacting with the world as it is and I'm not disparaging either side. I think both things are crucial, we need people that are thinking about things in both ways. But on my side, I find myself more interested in dealing with the world as it is. And as it is, I feel like selling the digital files of my music is, it's cute. There's this idea that I'm ascribing some sort of monetary value to my music, which, you know, maybe changes the perception of what is the value of this piece of art? …