Computer Music for Everyone: An Interview with Madrona Labs

Randy Jones on Madrona Labs History and New Instrument Sumu

Eldar Tagi + Ryan Gaston · 12/04/23

We at Perfect Circuit are heavily entrenched in the world of electronic music hardware. If it has knobs, faders, or heck, pressure sensitive keys? We're all completely on board.

But, that's not the whole story. Many of us at PC—and especially on the Signal writing staff—are also computer musicians. We're quite happy to work in the computer alone...using various programming languages to make our own software from scratch, or just working in a DAW as a primary part of our musical process. We don't shy away from computers, and in fact, we celebrate them for everything that they're uniquely good at. Instant recall, complex algorithms, nearly endless user customization, user interfaces that might be impractical if ported to hardware—there are plenty of reasons to favor computers over similar hardware instruments.

One of the biggest reasons, though? Computers offer a platform for creative coders to create and distribute groundbreaking instruments that just might not be possible or practical in hardware formats. The relative ease, low cost, and speed of software development means that game-changing tools can evolve at a surprising pace—and as such, we often see truly new developments in synthesis, processing, and DSP in general emerge first in software form before the world of hardware can catch up. Because of considerably lower R&D costs, software developers can take bigger risks than hardware manufacturers. As a result, some of the most unique, daring, and fascinating tools exist only in software format.

Madrona Labs is a prime example of a software-focused company bringing new sonic possibilities to the masses. Primarily the work of developer Randy Jones, Madrona Labs has consistently produced thoughtful, deep, and surprisingly intuitive means of exploring new sonic territory. His designs are consistently ahead of the curve. Take Soundplane, for instance: a multi-touch expressive controller that predates the MPE standard. Then, there's Aalto—one of the first widely-distributed software instruments that looked to Don Buchla's work as a source of inspiration, before the current "West Coast" craze took hold. And then, there's Kaivo, a forward-thinking software instrument that combines elements of granular synthesis and physical modeling into a novel new approach to sound design. We could go on, but the point we're making is that Jones's work looks consistently to the future.

His latest endeavor is the soon-to-be-released soft synth Sumu, which is based around the concept of spectral analysis and spectral decomposition. It allows you to analyze sounds and break them down into spectral components. You can think of it somewhat like granular synthesis, but instead of crafting sound purely from small "horizontal" segments of time, you also have access to the individual "vertical" partials that make up a sound...and you can do quite a lot to alter how these partials play out relative to one another over time. The end result? Sound design based on time-stretching, time compression, and frequency-dependent temporal manipulation. These techniques have long been exclusive to high-end DSP systems like Kyma, or to musicians who were capable of building their own software from scratch...but soon, they'll be available to everyone, with a simple and intuitive modular-style UI similar to Aalto, Kaivo, and Virta. In fact, this type of accessibility is key to Jones's work: it's all about taking novel new instrument concepts and putting them in the hands of artists.

We recently had the chance to talk with Jones about the history and future of Madrona Labs, with a special focus on Sumu. Read on to see what he had to say, and of course—be sure to check out the Madrona Labs website to learn more, and to experience his work for yourself.

An Interview with Randy Jones

Eldar Tagi: Hi Randy, it's a pleasure to have this conversation with you. Could you take us on a journey through your career, perhaps outlining the key milestones that led you to the establishment of Madrona Labs?

Randy Jones: I guess I'll start with the first sort of related job that I had after I moved to Seattle and that was making video games. The company I was working for at the time, Zombie, was making some pretty interesting games, including a futuristic racing game called Locus where you drove motorcycles on the inside of increasingly elaborate geometric surfaces. I was excited about the potential of video games to evolve into an art form and being a part of that, but it was just a couple years afterwards that they pivoted to sort of Call of Duty-style war games which was not even something I could be around.

Shortly afterward I was able to find work, making tools for creative expression, which suited me a lot better. I worked on a thing called Onadime, a sort of video synthesizer that got used by a few artists, but which never got a proper release. After that I was making my own software for doing live audio visuals, which was just becoming possible at the time with portable graphics hardware, and Cycling74 was looking for a solution for people to do the same thing as part of what eventually became their Jitter software. So I had something already underway, and I was kind of in the right place at the right time with it.

After working on Jitter, I went back to school to get my Masters in computer science, where I had a great time doing research but also discovered that a teaching career was not the right next step for me. This was in 2008 when the economy was crashing, so it was a good time to start something on my own. I took the Soundplane project I’d done for my thesis work, and got some friends together and said let's actually start making these, and that led to everything else I'm doing at Madrona Labs.

[Above: video demo of Madrona Labs Aalto, a virtual instrument inspired by Don Buchla's instrument designs.]

ET: Madrona Labs instruments are very distinctive both aesthetically and functionally. Can you outline for us your personal philosophy toward synthesizer design?

RJ: Well, I'm glad to hear you say that because it's hard to judge your own work and I'm not sure how much they stand out. One thing I'm obviously pretty adamant about is simplicity. This manifests in the one-page design of all the plug-ins. With a small number of exceptions, which are easy to categorize, you can see the complete state of the instrument without moving through different menus or pages. And I think that's important.

Too often, music programs are not humanely designed in that they impose on us unnecessary cognitive load. They make us remember where we are in a map, or even make us read through what are essentially spreadsheets to find, for example, our modulation mappings. With better design, we can reduce the cognitive load and make getting into a real creative flow state more of a likelihood.

ET: As an instrument designer yourself, there must be people or experiences that have been critical in defining your creative vision. Could you share who and what these influences are and how they've shaped your work?

RJ: Well right away Don Buchla comes to mind—I mean, he had the opportunity to start with a clean slate and develop his own ideas about what electronic musical instruments would be in a very special time and place, for special collaborators, and the resulting work was pivotal in our field. I'm a big fan of making what we could say are natural correspondences in audio and in perception a big part of the sonic palette of an instrument—and Don's low pass gate is definitely an example of this.

Also, there's a whole lineage of people that worked toward the kind of multi-touch, pressure-sensitive controller idea, and out of these Lippold Haken has had the biggest influence on me. The commitment to longevity, community, and support shared by Haken Audio and by the whole Kyma community is inspiring.

And of course, the same thing could be said about the monome folks. Kelli and Brian were super supportive and forthcoming with their own experiences when I started making hardware, and what they were doing was always a big influence.

ET: Madrona Labs instruments are very intuitive, and it seems that you have made several design choices (e.g. omitting menus, dial visualization) specifically to make them such. Can you elaborate on your approach to striking this balance between sonic complexity and user accessibility?

RJ: One way to keep a system simple but allow complex and surprising outputs is to enable its sub-modules to interrelate in more ways than just one so, for example by turning up the resonance on the filter, you can have another oscillator or by looping an envelope you can have a new LFO. And I always look for opportunities to do this.

I think the best way to make a system accessible has to involve a deep consideration of the medium that you're designing it in. Obviously, skeuomorphism is not the best approach, as most people have realized now. But we should be looking more fundamentally at what are the most ergonomic gestures and what are the most readable combinations of pixels to control with them. I'm just starting down my own path of finding new and intuitive designs with things like my animated dials, for example, but I know there's a lot of unexplored territory out there and ways that we can specifically use the computer to make transparent and usable and delightful interfaces, and that's something I'm excited about.

ET: How often do you get surprised by how users approach the instruments you've designed? What are some of the memorable examples of that?

RJ: Bob Ostertag came to town and played for us in a train yard with a game controller, Max/MSP, some portable speakers, and Aalto doing all of the sounds—that was super memorable!

[Above: official product demo of Kaivo, the Madrona Labs physical modeling + granular synthesis instrument.]

ET: You mentioned Don Buchla earlier—of course, the influence of Buchla instruments on Aalto is evident to many users. But with your other instruments like Kaivo and Virta, pinning down a concrete inspiration source is not as straightforward. Can you talk about where these designs come from? Were there any particular existing instrument references?

Screenshot of Madrona Labs Virta Screenshot of Madrona Labs Virta

RJ: I read a lot of academic papers on new methods of synthesis and control, and also as a small company that doesn't have a larger staff to do work like making 10,000 patches, the best way I have to make things that people will want to buy is to do things fundamentally different from other folks. So you won’t find me doing, for example, a wavetable synthesizer, or emulating some analog hardware. In the case of Kaivo, the really new thing was the ability to do two-dimensional (FDTD) models in real-time. And then the quality of physical modeling sounds which is in some ways very alive, but also lacks a certain dirt or grunge that lead me to feed it with a granulator. In physical papers they call this the residual, the part the model didn’t capture, so I said fine, let’s start with something very noisy and residual-like.

Virta is also based on some specific ideas from the past and one of these is definitely from the Korg MS-20’s audio-to-frequency module, which is also sort of present in their, little and lesser-known X-911 guitar synthesizer. I used to have both these devices and had a ton of fun mangling every kind of instrument through them and did not know any way to do this with softsynths. Then this led me to think about what the whole instrument would be like if it were aimed at making cool vocal sounds and didn’t have the standard subtractive layout. A patchable vocal resynthesizer / mangler.

ET: So, let's talk about Sumu. We're all excited to hear more about this new instrument of yours. Could you summarize for our audience what is Sumu, and how you envision it augmenting the existing range of Madrona Labs devices? As an extension of that question, I am curious about this combination of additive and FM synthesis. In the traditional additive method, a complex timbre is derived from blending together sine wave partials of different amplitudes, but in Sumu each of the 64 partials is an FM operator that already is capable of timbres of a certain complexity. How did you land on this voice architecture?

RJ: Sumu is a combination of additive with FM synthesis that we will release this fall. It's not quite done, but it's already making really interesting sounds and I'm getting better at explaining things and I kind of have my elevator pitch for it.

[Above: screen shot of the Sumu beta release.]

FM synthesis works great with sine waves or a few other simple waves, but if you take a whole sample and use it in FM, either as carrier or modulator, because of the way FM creates new partials you quickly get just a bunch of noise. There are too many partials generated and the results are not pretty, or what you'd expect. But if you take that sample, decompose it into sine wave partials, and modulate each of them with another sine wave signal you get the sound you would intuitively expect from frequency modulating a sample! it just sounds really cool and I like to think it's a new tool in the tradition of West Coast synthesis. Like the (Buchla) complex oscillator adds harmonics that you can dial in, we’ll be able to do that with any sample.

Plus, it's got this very general-purpose modulation source called Pulses that lets you move between periodic and random modulations very fluidly. And it's got this Space module that lets you zoom each partial around individually in a vector field…I'm excited to get it out there so I can focus on making some music with it.

ET: How would you describe the sonic potential, and perhaps the sonic character Sumu?

RJ: I'm tempted to say it's digital-sounding because it doesn't sound like overdriven tubes or transistors, but most often we associate digital sounding with the artifacts of the older digital equipment that you can hear, like quantization or aliasing. Whereas now, digital computers just produce sounds that are completely neutral if you want that. So it really attempts to make a very clear and natural sound as a starting point. Its character is more about the relationships between partials and the spatial movements that you can set up by patching it.

ET: The anticipation for Sumu has been building over the years. Undoubtedly, developing such a complex instrument comes with its set of challenges. Could you take us through this journey, sharing some of the hurdles you've faced and how you've overcome them?

RJ: Oh man, it's definitely been a longer journey than I planned. I put down most of Sumu’s design on paper in early 2019 and most of it is completely unchanged from that design. One thing I saw right away is that in order to make it work I'd have to do a bunch of big animations.

Now, the software library I had been licensing to do animations for my previous plug-ins was already stretched to its limit in some ways, for example, drawing Kaivo's animated dials on a Retina display. And I knew it wouldn't stretch far enough to let me do Sumu, So I thought OK, I'll make my own graphics framework for Mac and Windows and Linux, and this should take me maybe a couple of months—ha ha. Well, between that and getting Aaltoverb out and keeping the other plug-ins updated, a few years have gone by.

[Above: an assortment demonstrations by Madrona Labs using the Sumu beta release.]

I’m tempted to blame Apple for introducing a new processor architecture along the way as well, but honestly the main culprit was my tendency to go deep into writing code as almost a soothing activity, to really perfect things that I had in mind, see them through 100%, and that is not any sane way to run a business.

I've been spending more time on promotion lately and I'm excited to grow the team a little bit to help with this and get some of my more ambitious ideas done. Working as a solo founder is difficult for most people, so I should probably not judge myself too harshly and instead celebrate that I’m still around.

Finally, they don’t tell you this but programming is an endurance sport. When I’m super into it and the sounds are exciting, I work long hours and have trouble taking breaks, and I’ve had issues with RSI as a result. When it gets acute the only thing I can do is step away from the keyboard for a week or more. So I don’t know who needs to hear this, but: Take regular breaks! Hydrate! Stretch!

ET: On the Madrona Labs website, Sumu is presented as the final addition to the "patcher-in-the-middle" family of instruments. Could you explain what led to this decision?

RJ: Well, I think four is a nice number of instruments to build on one basic model and honestly I'm just ready for something new. These are tools for other people to be creative with, but they're also creative things in their own right. You don't want your favorite band making the same album over and over! I mean lots of people do want that actually, but it's no fun for the band.

ET: Given our interest in hardware, we're curious about the idea of using hardware controllers with your virtual instruments. Have you found any particularly interesting controllers or control strategies for interfacing with your software? Do you have any particularly fun/gratifying mapping strategies you can share?

RJ: One of my favorite things to do is set the envelope to retrigger and set the speed of that re-triggering to polyphonic key pressure like you can with MPE. You can either leave this un-quantized, or quantize it to get totally different kinds of rhythms, and I find it very satisfying to play. It kind of lends itself to grooving on Morton Subotnick-style rhythms. You can do this on a Soundplane, a Linnstrument, a (KMI) K-board, on whatever you want that has poly pressure—and it’s great there are some different options out there now.

[Above: the Soundplane Model A; image via Madrona Labs website.]

ET: We're also curious, of course, about the progress on the Soundplane Model B. Could you share any updates on this project?

RJ: Well, I talked a little about growing the business, and I think to do a sane job of supporting hardware I need to do that. They say that Luck takes care of fools or something like that and that was definitely true with the original release of the Soundplane.

Knowing what I know now, I would have more spare parts for example and more resources for teaching people how to use them and maintain them. So while I'm still very interested in making a Soundplane Model B, and there's actually a ton of work done on it already, I don't have anything like a fixed timetable for it. It completely depends on the growth of the company as a whole.

ET: With your wealth of experience and unique insights into the industry, I'd like to conclude our conversation on a visionary note. How do you envision the future trajectory of music technology? What innovative paths would you be excited to see this field explore?

RJ: There are some really simple visions that haven't really been pursued—partly this is because simplicity is hard. One of these is making a kind of musical appliance, or audio production appliance, and the best parallel I can think of for what I want to do would be the Canon Cat, a word processor designed by Jef Raskin in the '80s. The idea with the Cat was to make a useful tool that would get out of your way and let you work in a particular problem space. It was incredibly visionary for its time and had features we still don’t have—like if the power failed at any point it would have all your work saved. And principles of humane design were embodied in the thing at a deep level. I think there is potential for a kind of tool or appliance that could allow people to work with audio delightfully and elegantly—everything from simple recording to composition.

I think there's also a lot of room for new tools for people who don't really consider themselves composers or audio pros, tools with a lower barrier to entry that can enable new ways of creating music. For a long time, there's been a vision in the Computer Music field of easy access for the beginner while leaving headroom for the expert. There are so many interesting things to work on, there. And any use of computers that makes creative focus easier and more widespread and doesn't pull you into the social media vortex is a good thing to put out into the world.

How to Learn More

Talking with Jones was a distinct pleasure—we have a strong admiration for his work, and we hope that you'll take the time to check it out for yourself.

Sumu is due to be released quite soon: sign up for notifications about its release on the Madrona Labs website. And while you're there, do check out Aalto, Kaivo, Virta, and Aaltoverb—these tools have found their way into a huge number of our own musical productions, and we suspect that you'll find them as fascinating and inspiring as we do.