Interview with designers: Émilie Gillet (Mutable Instruments)
Among the Eurorack makers around the world, Mutable Instruments has become very popular immediately as a pioneer of ideas that make full use of the advantages of digital and development styles that open circuits and cords. We interviewed the representative, Émilie Gillet, by sending and receiving an email. It was a very deep topic, such as how you started the manufacturer, the design policy, and the future of Mutable and Eurorack, but I think it was an interview that you can understand his thoughts well. Then please!
Can you tell us how Mutable Instruments started? What were you doing before you started Mutable?
Before starting Mutable Instruments (MI), I was interested in signal processing (especially automatic song detection such as tempo extraction, genre recognition, etc.) and machine learning. I worked on these technologies with Google, Last.fm, and a small French startup. My main work was research and the content of research was put into software.
I started playing with Arduino boards in 2009 to study electronics and embedded systems myself. My first serious electronics project was a monophonic hybrid synth. Some people wanted to talk about the synth on several forums, so I decided to start selling it as a DIY kit. This kit (Shruthi-1) became so popular that after many times it was sold out, I had to set up a company to deal with taxes. And this company has become my main activity and main source of income.
MI originally sold desktop synthesizers in the form of DIY kits. They are Shruthi, a hybrid monosynth, Ambika, a hybrid polysynth, and Anushri, an analog mono. Why did you leave the DIY world? I was frustrated with the ever-increasing support and wanted to do something more powerful and beautiful.
When did you first come across modular synths (especially Eurorack) and when you first knew, did you want to do your own Eurorack maker right away and why?
I've often received requests for Shruthi's oscillator module, but Eurorack was strange and unclear to me. I needed an LFO, a clock, and a standard oscillator waveform to test the circuit I was designing, so I bought a 2012U Doepfer system in 6. In other words, I bought it as a test equipment. But I was immediately captivated and decided to make a module in this format. I was able to solve a problem I was running into at the time-because I was designing a big polysynth (like the high-end version of Ambika) at the time, but was it risky to mass-produce it? I was at a loss. It was the first big investment project for me. However, the Eurorack module seemed easier to get started with, it developed more and more, and the production costs were reasonably affordable.
Sure, your products were open source, open hardware (source code and schematics were released and could be reused if the conditions were met) from the beginning. Why did you do that?
It was because I was in the world of science and academics that I take it for granted. I was surprised that the hardware and plug-in companies were promoting their products as if they contained magical ingredients. I wanted to dispel the anxiety about such a way of revealing the contents. It's also very frustrating to run into internal firmware and software restrictions when using hardware. We wanted the user to be able to customize the module to get around those restrictions.
Your open source strategy has been a great success.ParasitesProduces a third party firmware likeOrnament & CrimeIt has been reused in other open source modules like. Did you think that your way would affect other designers so much?
No. I didn't expect to see anything deep like Parasites that could be called "official unofficial firmware". I expected 10 or 20 unofficial firmware with minor changes and improvements to suit individual needs. There's a gap between what the other designers thought they would copy from my module and what actually happened. I was expecting something a bit more dialogue. Looking at my schematics and code, finding flaws and suggesting better ways.
But I'm really happy to see something like Ornament & Crime.
Did you publish the code did not lead to debugging or improvement? That's what I first imagine about the benefits of publishing code.
I am very careful in coding and have very few bugs when the module is released. What sets me apart from many developers in the musical instrument industry is that I write most of the code in a module in a cross-platform way, which allows me to test the core functionality of a module on my computer. I will. Assign random values to all knobs and CVs to see if it's okay if you keep them moving for hours, or check for strange situations that could take minutes to patch with real hardware. can do.
Perhaps more people were more interested in (more boldly) adding something to the code than just reading it, understanding it, and figuring it out for improvement.
Please tell me how to work in the office. Is there anyone else to help?
We have a partnership with a factory in Normandy, which handles production, testing and packaging. I will send it to me in the box of the finished product that everyone will have. At times, we will increase the number of people to 15 and get them to finish.
I myself do not hire anyone and do all the rest. My day job is divided into two, 1/2, which is business operation (email and shipping work with customers, module repair, business matters) and new product design. The only MI product I haven't done is graphic design and panel layout, which is also handled by other manufacturers.Hannes PasqualiniIt is by.
Do you do a lot of research into new synthesis technologies and algorithms, and do you think there are still synthesis frontiers that are not realized in the Eurorack world?
I try to be sensitive to new research. Read through conference proceedings such as DAFX and ICMC, or read anything about machine learning. I'm working on a small collection of code that implements the ideas I encountered in the paper, and I think I can make about 20 modules with it. But to be fair, it doesn't work for Eurorack because of lack of computational power, latency, lack of fast modulation, or simply not being modular, but there are a lot of interesting technical topics. . And I think it's not just the futuristic things that are cutting-edge, but just the topics that are present. I still find many discoveries in the classic topic of interesting LFO waveforms and musically usable random data. Do some interesting discoveries such as "what kind of knob do you need for the filter module?", "What kind of waveform do you need for the oscillator?", "What's between the envelope generator and the sequencer"? Is enough for
Do you have a Eurorack setup to make your own music, what modules do manufacturers other than Mutable have?
I had a small system with Doepfer, Intellijel and Make Noise, but every time I used it, I felt like I was testing or debugging, and I wanted to modify the module, so it was a personal use. It wasn't. I moved to a new house last October, and at that time I decided to make it a place for reading and listening to music, and avoiding technical equipment as much as possible. I became a type of person who did not have a modular in the house like Dieter (Dieter Doepfer, founder of Doepfer). I have a small setup of 10 or 4 modules in the workplace, all of which are under development by me. It's not about playing music, it's about testing and playing around.
As seen in INA / GRM, France has a long and full history of electronic music and music concrete. Do you think you are influenced by such history and research?
Although I have not received it directly, I have to be aware that such things were always close to me. For example, I remember that a junior high school textbook had a chapter on electronic music, and I was shocked by the photos in the studio. If you know that such a thing exists in a proper and serious research institute, it will be the goal of your career. France has traditionally been strong in signal processing and mathematics education, and has personally benefited from it, and thanks to this, there are also institutes like GRM and IRCAM.
Digital modules can be overly complex, requiring menu navigation, one knob having multiple functions, and firmware updates. Is there a design policy for this complexity? Will Mutable ever come out with a module made up of a large OLED screen and encoder?
I've made a number of mistakes about it. It would look like this if we were to make design guidelines now.
-In any situation, the label on the panel and the function of the knob must match.
For example, don't make a collection of extraneous features like Peaks (knobs simply labeled 1,2,3,4), Clouds' hidden settings, Easter Eggs, official firmware updates, etc. However, variations like the Rings mode are fine.
-The module allows the musician to infer settings as automatically as possible depending on the state of the patch, rather than letting the musician press a button.
Like Rings, the idea is to switch the trigger of the sound inside or outside depending on whether it is patched to the signal input or not. (Almost unnoticed, this is innovation!)
-The combination of OLED and encoder can be used if it is of a type such as "set once and set aside", such as a MIDI interface.
IER-301I can't think of designing something like.
What comes to my mind right now is something like "convexity of a parameter". A good module should be able to result in an A in one setting and a B in one setting, and any result in between. You should also avoid going into areas where you find yourself not able to do what your module can do as you pursue more and more of what your module can do. With such a design principle, for example, a module with unrelated functions, a module with a bonus function that CV cannot control much, etc. will not be accepted.
Tides and Rings are my favorite modules and templates for my design right now. I don't like Braids very much and I think it was a mistake for Clouds.
Could you explain a little more about "how to bend"? To me, "I can get any result between A and B" is not "how to bend" but "continuity". Does it mean that you have to adjust the curve of the curve connecting A and B properly so that there are as many musical sweet spots (points at which good sound is produced) as possible in that curve?
It's not just continuity. It's important to be able to transition between the two settings in a continuous path that keeps users from getting into unwelcome parameter areas. In a simple example, an oscillator that can output a pulse wave with a width of 2% to 1% by adjusting the pulse width. This allows a smooth transition between two settings from 99% to 1%. This is a continuous path, but it also leaves some areas unreachable (like more ringing sounds). You can't solve the problem simply by moving continuously with the knob.
That's right about the sweet spot. That's a key concept. You should aim for the sweet spot when you turn the knob.
The "random set of features" found in Braids and Clouds seems to be favored by users, unlike your tastes. Are you worried about this mismatch?
Not really. First of all, Braids and Clouds became popular because there was no similar one in the Eurorack market. Even if Clouds has only the main granular mode, I think users like it. Rather, they may have spent more time learning how to use it and liked it more. Braids are the same. Even if you turn on the power and don't touch the encoder, it's a nice oscillator. I've really wondered if there's a “Model of the day” feature that will randomly select one mode when you turn it on and will lock into that mode until you turn it on again.
Also, if users like the old, feature-packed modules over the more consistent modules that I have in my mind, I'm happy to personalize what I like. .
Mutable for 2017 is still pretty static, can you tell us about your plans for 2017?
There are new modules that complete the current line-up, and there are modules that have been redone from the root. It's not an upgrade like Mk II, but an idea like this. `` What is the new module X, Y, Z that can do the same thing by patching the existing A, B, C modules, and also considering the bending more? ''
That's an interesting way. In the new framework, which is conscious of how the parameters move, the new module will be a more coherent set of functions like Tides and Rings rather than a set of unrelated random functions, right?
That's right. It doesn't mean removing multifunction modules, but it does mean that every function is derived from one underlying principle, or at least the roles of knobs and jacks are uniform. Tides, for example, act as LFOs, VCOs, and envelopes, but this is not an NG. The reason is that looking at the contents of the module, the same source code raises and lowers the voltage in all modes, and the various functions described above only raise and lower once (envelope) The only difference is whether to repeat at low speed (LFO) or repeat at high speed (VCO). Serge's DSG is a module with similar functionality, allowing the underlying elements (voltage up and voltage down) to work differently.
Can you expect to find the answer to the question "what's between the envelope generator and the sequencer" in a module designed with your new framework?
Yes. But don't get mad if it doesn't get released in the next six months, or if it doesn't.
What do you think the Eurorack world will look like 10 years from now, and do you think more interesting equipment and frameworks are emerging than Eurorack?
We are confident that in 10 years, half of our current users are moving to something else. However, new users will come in that much. As you get older, you become nostalgic, and some young people will want to do that.
What was the world of synths like in 2007? With lots of plugins and desktop virtual analog synths, there weren't many real analogs, and when it came to modular, it felt like confidential information. It's not surprising that there will be even bigger changes than those from 2017 to XNUMX.
Is there a better framework? Perhaps there are better modular formats than Eurorack, and even better for those that are not oriented towards CV control. Along with Eurorack, there is something that I call the "tabletop orchestra", which is growing. Volca, Pocket Operator, Meeblip, and guitar pedal ... It's a style that connects these on the table. There may be opportunities to integrate with Eurorack. In that case, desktop units and modules have the same standardized size, power connector, port, etc., so that they can be put in the same case. I'm careful about tools that have been extremely integrated from the beginning. If you connect various things that are not directly related to each other, it will be twisted, glazed, and even the reliability will be impaired, but this twist is imaginative and not a defect.
Selected Modules from Mutable Instruments
¥ 35,300 (Tax excluded / tax excl)7-output random CV / gate module that can control temporal correlation and distribution shape
MUSICAL FEATURES Marbles is a random gate / CV generator with many outputs and CV inputs.The output random voltage can be limited in various ways (eg, synchronization with an external clock, frequency of repetition, appearance of rare events, traditional stepped random voltage, etc.). t ...details
¥ 26,300 (Tax excluded / tax excl)Oscillator-synth voice module with 16 models
MUSICAL FEATURES Plaits is a digital oscillator / synthesizer voice module that can use many models (algorithms). Mutable's old oscillator Braids design has not been inherited, and both hardware and software have been redesigned from scratch. Many algorithms ...details
¥ 11,500 (Tax excluded / tax excl)External input module with contact microphone
MUSICAL FEATURES "Ears" is an external signal input module with a contact microphone, which was collaborated by module maker Music Thing centering on DIY kit and Mutable Instruments. Music Thing's Mikrophonie ...details
¥ 35,300 (Tax excluded / tax excl)A multi-modulator that combines 6 stages to create multiple envelopes, LFOs, and sequences
MUSICAL FEATURES Stages is a modulator that combines multiple stages to create multiple envelopes, LFOs, and sequencers. A complex 6-stage envelope, one AD envelope, and one 6-step sequencer are also possible. How to combine the stages ...details
First of all, could you explain how Mutable Instruments started? What did you do before you started your company?
Before starting Mutable Instruments, my two main interests were signal processing (especially automatic analysis of songs-to extract the tempo, recognize the genre, etc) and machine learning.I worked on these kinds of things in large tech companies: Google, Last. fm, and a rather obscure French startup. My job was mostly research, and turning research into software.
I started playing with Arduino boards in 2009 to teach myself electronics and embedded systems.My first serious electronics project was a hybrid monosynth.I talked about it on a couple of forums, and since other people wanted to build one for themselves too I started selling This as a DIY kit.This DIY kit (the Shruthi-1) became quite popular, and after having sold several batches of them I had to create a company for tax reasons, and this company gradually became my main activity and main source of income. .
Mutable Instruments was originally selling desktop synths in the form of DIY kits: the Shruthi (a hybrid mono), the Ambika (a hybrid poly), and the Anrushi (an analog mono) .What drove me away from the DIY world? The increasing frustration with support issues and the need to do more powerful and beautiful things ...
When did you come across modular synthesizers (especially Eurorack)? Did you decide to become a Eurorack manufacturer yourself soon after this first encounter? What made you go in that direction?
People had been asking me for a Shruthi oscillator module for a while but Eurorack was really something strange and foreign to me. In 2012 I bought a 6U Doepfer system, primarily because I needed LFOs, clocks, raw waveforms etc to test the circuits I was designing, like the Shruthi filter boards.I really thought of it as a purchase of test equipment! But I instantly got hooked and decided to make modules in this format.It solved nicely a problem I was facing at the time; I had designed a big polysynth (a higher-end version of the Ambika) and I was hesitant as to whether I should take the risk to launch it via mass-production.The money I had to invest was very big, and this was my first serious industrialized project. The Eurorack modules looked like an easier way to get started, with a more progressive curve and production costs I could afford.
If I remember correctly, your products have been open-source and open-hardware from the start.Why did you make that decision? (Related question follows)
It's part of my scientific / academic background in which sharing is a habit.It's because I'm really appalled by the way some hardware or plug-in companies make people believe there's a "magic ingredient" in their product: being transparent about how your product works is a remedy against that.It's also because I know how frustrating it can be when we hit firmware or software limitations when using a hardware product-I wanted to give my customers the ability to customize their modules so they could bypass these limitations.
It seems your open source strategy has been a great success, because many third-party alternative firmwares have emerged like Parasites, and your codes are recycled in other open-source modules like Ornament & Crime. Did you imagine your strategy would have such great influence on other designers?
I did not really anticipate what happened.I did not expect something like Parasites, which is very deep, becoming the "official unofficial firmware". I thought there would be ten or twenty alternative firmwares, with minor changes or improvements suiting personal needs.
There's a bit of a mismatch between the things I expected other designers to "copy" from my modules, and what has actually happened. Somehow I expected more of a dialogue --people looking at my schematics or code and finding flaws, giving me suggestions about things I could do better. But I'm really glad to see something like the Ornament & Crime.
So making your code public did not help you debug / improve your code? That's what I would have expected to be the chief merit of making the code public in the first place!
I'm very meticulous with coding and there have been very few occurrences of bugs getting into the modules once they are released.One thing that sets me apart from other developers in the world of music gear is that a large section of each module's code is written in a cross-platform way-which means that I can test and run the core functionality of the module on my development computer, not on the actual hardware.This allows me to conduct “monkey tests” in which I feed random values to all knobs / CVs for hours and check that nothing crashes, and more generally to check what would happen in strange situations that would take minutes to reproduce with the actual hardware.
Overall I think there are more people interested in adding to the code than just reading it and trying to figure out how it works and how it could be improved.
Could you explain how you run your business at Mutable HQ? Do other people help you there?
I have a partnership with a contract manufacturer in Normandy who handles all the manufacturing, testing, packaging-they ship me palettes of finished boxes, the same boxes you buy.At times there are up to 15 people working simultaneously on the modules in their factory As a result, I don't have any employees and do everything else by myself. My day is split 50/50 between running the business (answering customers' questions, shipping orders, repairing modules, administrative duties), and designing new products. .The only thing in a Mutable Instruments product that I do not create myself is the graphic design and panel layout-this is handled by Hannes Pasqualini, who now designs for other brands too.
Do you often research new techniques / algorithms for synthesis? Do you think there are still many new frontiers in the field of synthesis that have not implemented yet in the world of Eurorack?
I try to stay up-to-date with the latest research, reading the proceedings of conferences like DAFX or ICMC plus whatever is going on in machine learning.I have a collection of small code snippets implementing ideas I encounter in papers, enough to make 20 modules or so. But to be fair, there are a lot of exciting things that are not ready for Eurorack, either due to issues of computational power or latency, or because they would be deemed too "unmodular" or would not handle fast modulations And the future is not the only frontier, we can also look at the present.I think there are still many things to be discovered on boring, old-fashioned, small things like generating families of interesting LFO waveforms or musically useful random data. Just questioning things like "What knobs should there be on a filter module?", "Which waveforms should be available in an oscillator?", "What's in the space between an envelope generator and a sequencer?" Is enough to enable interesting discoveries .
Do you have your own Eurorack set-up for music creation? If so, what other manufacturers / modules do you have in it?
I had a small system with Doepfer, Intellijel and MakeNoise modules; not my own modules because every time I used them it felt like doing testing / debugging, and this gave me urges to tweak them. Last October I moved to a new apartment and decided that it would be a space for reading and listening to music, with as little as possible technical equipment, cables, clutter ... so I did not bring any music / audio equipment in.I guess I'm like Dieter Doepfer now, with no modular at home.At work I have a small setup with 4 or 5 modules-everything new I'm developing; but it's more testing and playing around than a true musical project.
France has been famous for having long history of electronic music / musique concrète like INA / GRM Are you influenced at all from that kind of history / research?
Not directly, but I can't help thinking that this stuff has been around all the time-for example there was a chapter about it in my junior high school music textbook and I remember being impressed by the studio photos. Just knowing that this thing exists, that it's happening in a legitimate and serious institution is enough to turn it into a career goal.There's also a good tradition of teaching signal processing and mathematics in France that I benefited from, and that accounts for the existence of institutes like the GRM or IRCAM.
Sometimes digital modules get too complex because of menu-diving, multi-functions per knob, many firmware updates, and so on.Do you have any specific design policy on module complexity? Will we see modules from you in the future that have big OLED screens and encoders?
Well, I made lots of mistakes on that front! My current set of guidelines would be:
-Under any circumstance, the label printed on the panel should match the knob's function.
This prohibits modules that are collections of unrelated functions like Peaks (with knobs labeled simply 1, 2, 3, 4), hidden settings like Clouds, or easter eggs, and releasing "official" alternative firmwares. Variations (like Rings' various modes) are OK.
-The module should try to infer as many settings as possible from the way the module is patched, instead of asking the musician to press buttons.
The way Rings switches to an internal / external exciter depending on whether the signal input is patched is a key idea (an innovation few noticed!).
-OLED displays and encoders might be OK for something requiring many "set and forget" settings, like a MIDI interface.
But I don't see myself designing something like the Orthogonal Devices ER-301.
The notion that stays a lot in my head at the moment is that of "convexity". A good module should feel "convex" in the sense that if you can get result A from it and another result B from it, you should also be able to get anything inbetween.Or the things it can do should not push you towards a direction where you'll notice the things it cannot do.As a design principle, that rules out a lot of things, like collections of unrelated functions, or modules with "bonus" features that are not fully CV-controllable.
Tides and Rings are my favorite modules, and they are serving as templates for new designs at the moment. Braids is not my favorite module, and I consider Clouds a mistake.
Could you elaborate more on "convexity"? "Getting any result between A and B" sounds like continuity, not convexity.Do you mean that you need to fine-tune the convexity of the continuous virtual curve linking A and B, to make the curve have as many sweet spots as possible?
It's not just continuity (or path-connectedness: that would be the more accurate notion)-you can have a continuous path between two settings that still does not get you where you want it to go.A silly example would be a VCO whose PW control behaves this way: turn it CCW and you get a pulse with a 1% ratio, turn it CW and you get a pulse with a 99% ratio, and inbetween you just get a crossfade between these two settings.That's a continuous path, but it still leaves a gap, there's a type of sound (the “hollow” square with a 50% duty cycle) that this would not cover.So having all continuous settings (all knobs, no switch) doesn't solve the problem.
But you're right about the sweet spots-that's the key notion ... Make sure that when turning a knob, we visit the sweet spots.
People seem to love the "collection of random features" found on Braids and Clouds, as opposed to your like / dislike.Do you worry about this mismatch?
Not much. First, because what made Braids and Clouds popular was that there was nothing like them before on the Eurorack market.If Clouds only had its main granular mode I'm pretty sure that people would have still loved it-and maybe they would have taken even more time to learn how to use it and loved it more.Same for Braids-even if you power it on with one model and don't touch the encoder (I really thought about this “model of the day feature”- the module powers on in one randomly chosen model and stays locked there until you power it on again), it's a fine oscillator.
And well, even if people prefer the “old”, feature-loaded module instead of the more streamlined things that now occupy my mind-at least I will have the happiness of making things I'm more happy with.
Mutable Instruments in 2017 has been pretty quiet so far.What are your plans for 2017?
New modules completing the existing collection, but also reworking the fundamentals-not in the sense of MkII versions, but through this kind of approach: "Can we do what modules A, B, C do when they are patched together with new modules X, Y, Z-with X, Y, Z being more 'convex' than A, B, C? "
That type of reworking sounds interesting.So the new modules (serving in a more convex framework) won't have unrelated random functions, but will have more coherent collections of functions like in Rings and Tides?
Yes. This doesn't mean I will exclude multi-function modules, but if they are multi-function, it should be either from a same underlying principle, or at least with uniformity about what their knobs / jacks do. For example, even if Tides can work as an LFO, VCO, envelope generator, I don't see it as an “evil” multi-function module, because if you look inside the code of the module, you see the same code managing a voltage going up and down-and the various functions are just obtained by changing if this up / down movement is done once (envelope), or many times at low frequencies (LFO), or many times at high frequencies (VCO) .The Serge DSG is of this same kind of multi-functionality flavour-where there's one underlying principle that can be used and abused in different ways.
Can we expect modules in your new framework that will answer questions like, "What's in the space between an envelope generator and a sequencer" (ref answer to Q6, above)?
Yes, but don't nag me about it if it's not released in the next six months, or is not released at all.
What will the world of Eurorack look like in ten years' time? Do you think there will be a more interesting tool / framework than Eurorack at that point?
I'm pretty sure half of the people will have moved on to something else, but they'll be replaced by new people: old people getting nostalgic, young people wanting to get retro, etc.
How was the world of synths in 2007? Lots of plug-ins and desktop VAs, not much analog stuff, and modular was confidential.There's no reason 2027 will be less different from 2017 than 2017 is from 2007.
May be a modular format better than Eurorack, and not geared towards CV control.One trend that has grown in parallel with Eurorack is what I call "tabletop orchestras"-boxes like Volcas, Pocket Operators, Meeblips, guitar Maybe there would be an opportunity for merging this with Eurorack so that desktop units and modules could all have standardized sizes, power connectors, I / O ports and live happily together in the same case. I would be wary of some extremely integrated tool-there's a kind of "wonky" feel to connecting unrelated things together in a not-always-reliable way.This wonkiness is inspiring, and is not a defect that should be fixed.