Justin Coletti talks with Jonte Knif (Knif Audio founder and designer) about the Soma EQ, which was just launched as plugin. Also cross-genre mastering legend, Zino Mikorey shares his experiences with the Soma EQ. He owns and uses both hardware and software. Listen to the interview via SoundCloud or enjoy reading this blog article to get to know the process of designing and using the Soma EQ and all the thoughts and decisions it takes.
Justin Colletti: Today, we are talking to two amazing practitioners in their fields.
One is Jonte Knif, a designer of some fantastic, high-end, boutique analog hardware. In the past, Plugin Alliance modeled his multi-tube synthesizer, Knifonium, an absolutely gorgeous-sounding device.
Their most recent model of one of Jonte Knif's designs is the Knif Audio Soma EQ, a mastering-grade, passive tube EQ. We're going to talk about what all of those things mean and the benefits of the design.
We also have Zino Mikorey, who is an amazing accomplished mastering engineer. He works out of his own studio in Berlin and does projects at a really high level across genres. He is an owner and big fan of the original hardware, the Soma EQ, which he knows very well and loves very much.
We're going to get both of these guys' perspectives on this tool. We’re really excited to have this conversation about mastering and mastering EQs with them, with particular reference to the new Soma EQ modeled by Plugin Alliance.
All right, let's get right into it.
Jonte, as a designer, I want to get a sense of why you wanted to build a passive tube EQ. What are the benefits of that kind of design? And what do you think is different about it from any other tool like this on the market?
Jonte Knif: A fairly long time ago, I had some local customers who were very interested in having some sort of passive EQs. I was also personally very interested in the sound quality you can get from coil and capacitor-based passive circuits.
That's one part of the sound–the EQ circuits by themselves–but then there is the makeup gain amplifier, which I designed very carefully to be a sort of high-end tube amplifier, single-ended transformer-coupled, with extremely high-quality output transformers so that it wouldn't be too obvious or have too much character–it’s just the right amount of second harmonic.
Justin Colletti: So you're getting some of the character that you expect of passive tube-based, transformer-based EQs, but without going overboard. Some EQs of this style can be a little colored for some mastering applications, perhaps.
Jonte Knif: That's true. They might excel in mixing. They have a wider ranges. From the beginning, I was thinking more about mastering and that's why the ranges are only to 8 dB. I wanted to have a lot of frequencies available and also a real Q value adjustment, which other passive EQs don't have because it's fairly complex to build.
Justin Colletti: Great, before we get into Zino's thoughts on actually using this and how it's compared to other tools in his mastering work, can you give us a sense of what exactly the real Q adjustment means? To my understanding, this is unique and a first in the world of hardware passive tube EQs. What's the point of real Q adjustment? What does it do and how is that a differentiating factor for the EQ?
Jonte Knif: The traditional way to adjust Q in passive resonance circuits is to put some resistance in the path, which unfortunately, while widening the Q, also reduces the amount of cut or boost. So you can’t adjust the Q independently of gain in this traditional way. It has to be done with changing the capacitance and inductance values in the circuit. And that gets very, very complex compared to the traditional way.
Justin Colletti: So, my understanding, Jonte, would be that if you've picked one Q value—say a bandwidth of 1—and you do a 3 dB boost, and then you start to decrease the Q value, making the bandwidth even wider, all of a sudden you're not having the same level of EQ boost you had before. While it might say an EQ boost of 3 dB on your knob, you're actually having less EQ boost than that once you start widening the bandwidth. Is that how conventional passive EQs work and is that the problem you're trying to solve?
Jonte Knif: Yes. Yeah, that's the way.
Justin Colletti: Cool, now, I have a couple of questions for Zino. First off, why were you interested in having a passive tube EQ in your studio? Why is that useful? And is there anything particular that you think has been unique about the Soma in that style of processor?
Zino Mikorey: I didn't really want to use a passive EQ. I was just searching for an EQ to give me what I was searching for in sound. And I think the Soma has this sort of lens quality. You want to put a lens on something, you want to spotlight something and it can do it in this integrated way where it's not on top. It feels like you're just getting more of the original sound instead of something artificial on top.
And then the cool thing about its passive summing is that you can basically create your own curves. You can widen the curve with a flatter top. And that's the most interesting part for me about the Soma. The plugin brought the box tone back in my memory. At the end of the day, you can do so much, and you can create these very wide transfer curves that don't do too much to the face because it's so soft, so wide. And that's what I really like about it.
Justin Colletti: Are there any particular areas in the frequency spectrum where, if you have more than one EQ that you use in the studio, you say this is an area that I really want to touch with the Soma?
Zino Mikorey: Yeah, especially, let's say 600 Hz to 5 kHz is for me the Soma range. I'm not really into air boosting. For me, it's always in the way, if you boost some air, the noisy part comes to the front and distracts from the music that's in the midst. I tend to use it in the vocal ranges, the instrument ranges, the solo ranges.
I really like it for widening because the crosstalk is marvelous on the transformers. I used to use an EQ with a very low high-shelf to give the vocal a lift to the front. Now, most of the time, I think I'm doing this with the crosstalk when using the MS matrix. This is one of the parts that I really love about it. You get the widening but you also get the sort of structure and stability in the mids through the crosstalk.
Justin Colletti: Right. That's something I'd love to hear about first from you, Zino, a little bit on using this in MS mode, and then I'd love to ask you, Jonte, a couple questions about the design of it from an MS perspective.
Using this in MS mode, first of all, just give us your conceptual framework for using MS in an EQ to begin with, Zino. While mastering a track, what makes you think, “This calls for me to start using this in MS, rather than in stereo”? And what percentage of the time do you think you're working with it in Mid/Side mode rather than stereo?
Zino Mikorey: It's all about focal points. You have all the stuff that's in the middle and maybe your vocal is a little buried, so you need some 1K or whatever, where the vocal sits or maybe an octet higher or on octet lower. You want to create this focus in the mids–that's where I would use it, or for brightening up the sides and creating this sort of fake space.
It could be as crazy as the guitars needing 3 kHz on the sides because they're a little muddy, but the vocal is too bright and it needs meat to come to the front. So you cut in the middle and you boost in the middle to sort of fix the vocal, but then you brighten up the sides because the guitars are muddy.
I used to use this a lot. Now I use it very rarely because I do this in my mix feedback. I do all the big changes I do through feedback or at least I try to do it through feedback. I tend to not use it that much anymore.
It could work great. It’s as simple as boosting a little more of a frequency in the mids or boosting a little more on the sides and therefore shifting the focus, shifting the attention and creating a balance.
Most of the time, I really feel rather like a balancing engineer, even when mastering with the wide Q shaping. With the wide Q, you also don't have a lot of phase, so there's not a lot of time-shifting happening, or as I always also see it, a sort of time-masking as well.
On the other hand, if I want to unmask a vocal or if the song is a little too honky, midy, boxy, then you could boost where the boxiness ends, and this way, correct me if I'm wrong, Jonte, but everything sort of on top.
If I have a wide Q bell, everything on top of the corner frequency will be slightly earlier in time and everything underneath will be slightly later. So, now, I didn't cut into the honkiness and I didn't cut into the boxiness. I've just sort of shifted the boxiness back in time, and I shifted the octet above forward in time.
When the snare or vocal comes, I still have all the body, I didn't cut into it with a tight Q or create these weird time machine holes, but I still created this much clearer, much better and accessible, listenable signal. For me, most of the time, the Soma is just the best time machine possible because you can fix problems without cutting into the problem.
Justin Colletti: I love something you touched on there which gets away from the idea of EQ for a second, Zino, but the idea of earlier in your career doing more heavy-handed MS work, and now later in your career saying, "The best thing I can do for this client is give them a form of mix coaching or mix feedback after hearing it in my room."
I think that's an incredible value that real human mastering engineers bring that isn't talked about or noted quite as much. We think about how they made it sound before and after. But, if your mastering engineer can help make you better at mixing, my goodness, that's even more valuable than having them make your mix sound slightly better.
If they can help you improve the quality of the work you're putting out from your interaction with them, I feel like that's worth more than the price of mastering a track.
Zino Mikorey: You can't compare. The mix is much better than the master I could have built with cutting. And, back in the day, I used to butcher stuff, but in a very surgical, good way. People were always super happy. They called me the surgeon. And I loved it, obviously. But, then when you start getting into mix feedback, what you get out of it, or let's say you open up an old session and then you bypass a crazy dynamic EQ, you realize that you've taken away everything.
While you're in the surgical, listening-for-problems mode, you're just sort of butchering in the very best sense, going, "This is the answer. I'm such a great guy for doing this." Then, you come back and you listen and go like, "What the hell did I do?" I think at the very end, that's the relationships you make, everybody benefits. Everybody gets better.
And then in the very end, what every engineer should have, I think, is a healthy catalog. And this is the way people find you. They listen to your stuff and then go, "Whoa, everything he does sounds great." It involves a lot of work because you can’t make everything sound great with EQ.
Justin Colletti: The harder work is having that conversation with people and figuring out how to give them coaching in a way that they will hear and be excited to act on rather than feel like you're talking down to them. Those soft skills are super important.
I love that idea. By going really heavy-handed on a mastering job, you could improve their mix significantly and maybe make it 20% better, but if they went back and redid the mix, they could make their mix sound 35% better. Then, you're making it sound 5% or 10% better on that. And the end result is a better-sounding final product. So, I love that holistic approach to mastering that you're talking about.
Now, one of the things you also mentioned here, is bringing forward a vocal without bringing forward the sides, or perhaps brightening the sides to create more space.
To my understanding, Jonte, if you would talk a little bit about this, there is a fairly unique feature in this EQ that I'm not used to seeing in many EQs in the Mid/Side function, which is that there's two different ways of the low frequency roll-offs working the high pass filters.
Could you describe what you were going for with the different high pass filters for the Mid/Side mode and how that could be useful for people?
Jonte Knif: Yeah, so the high pass on the side channel has two options in the hardware and both channels in the software.
The 6 dBs per octave choice has much higher frequencies than the 12 dBs, which is really a bass cut, but this less steep option is not meant for bass cut, per se, but to make it mono.
It's the very traditional elliptic EQ, which has been around since the ‘50s or something. There is nothing really unique in that way–it's super traditional–but it's such a nice thing to have in an EQ. If there is a MS feature, then obviously there should be also this elliptical EQ.
Zino Mikorey: And, the cool thing is because it’s 6 dB, it's not so hard on the face. If you have it at 12 dB or 24 dB, you can get all the mismatch and–
Jonte Knif: Yeah, it's impossible to do this elliptical with steep filters, it just creates a mess.
Zino Mikorey: But, people do, and that's the problem. Somebody told me that on the crossover point of 12 dB or 24 dB it's out of phase, right?
Jonte Knif: 12 dB will create a 180-degree phase shift, not on the crossover point, but where the curve has already approached the steepness. That will create weird bumps to the mid signal, if that is used for the sides.
Zino Mikorey: And if you think about the whole encoding, decoding rate, it also means that now above the corner frequency, when it's out of phase, that means it's on the other side.
Jonte Knif: Yeah, true.
Justin Colletti: This is interesting because the idea of phase in EQ, correct me if I'm wrong, but if you're dealing with a single mono channel, the phase shift issues that people talk about in EQ, shouldn't be that big of a deal necessarily.
But when you're talking about EQ'ing two sides differently or a mid separate from the sides, that's where a lot of these cancellation issues create filter shapes that are very different than what you expect from occurring.
Is that correct, Jonte, that this is more of a concern in Mid/Side mode or in dual stereo mode than it would be just on a straight mono EQ?
Jonte Knif: Doing different EQ for left and right in stereo mode is obviously a big problem. If you just try to cut some problematic frequency from only one channel and you do it with a lot of gain and high Q value, there will be local phase problems, which will then probably screw things up quite badly.
In M/S mode, you are not tilting the stereo image so much, but the problems are the same. I'm not a mixing or mastering engineer, so I don't know what, in practice, these things cause.
Justin Colletti: Right, there's probably a degree in which you can get away with subtle boosts and cuts separately–left, right–but a point past which you're going to create local phase cancellation that could give you imaging issues and potential unexpected peaks and nulls near the point where you–
Jonte Knif: Yeah, I would imagine that the imaging goes really bad, quite fast.
Justin Colletti: Do you have any best practices on this, Zino, about how far you're willing to take separate processing or is it really just listening and deciding?
Zino Mikorey: The good old bypass trick is always the most valuable thing. Level-match it and see, close your eyes, click the bypass button as many times so you have no idea which is which, and then really decide.
But, again, I'm the same. I don't tend to overuse it. If I do it, I'm double-blind testing myself, but it also depends. For example, I might get a piano record where the piano on the left side is rather dark and the right side is bright-ish. Then, these are two mono sources for me.
So, if I EQ the left side, this doesn't necessarily have to be a bad thing for the mid because there's no remit, it's an AB, you know what I mean?
Justin Colletti: So, the more dissimilar the two sides are, the less of an–
Zino Mikorey: Exactly. I know of many tricksters that do this–a different Q on the left, a different frequency on the left, and different frequency on your right to create this fake widening.
I think if you use a super wide Q, passive design, or very clean design, you can get away with it. The worse the circuit, the more horrible it is. I rarely do it. If I do it, then I'm telling them, "Your mix is to the right." Or you can see it. That's also a very interesting thing, sometimes you see it at the mixing desk. I can just easily go in and perfectly mid-mono it so that I don't mono it obviously, but I'm boosting on the left side so the frequency's back in the middle.
So, most of the time when I do it, either I do a blind test the next day before I send it or I send two versions or communicate about it. I don't think in these modern times where you always have a recall, rarely people are really mixing on discs. The problem is better addressed somewhere else, or the opportunity is better addressed somewhere else.
Justin Colletti: And just to close out the whole idea of a phase and treating different channels differently, Jonte, based on what you're saying, it sounds like potentially you can get away with a little bit more in Mid/Side mode because the side channels are so dissimilar to the mid channel compared to the conventional way of doing different things on the left and right. Is that correct? Because your sides are so different from your center, you could maybe get away with a little bit more EQ than independent left, right EQ?
Jonte Knif: Yeah, I think it should be like that. The problem when working in stereo mode with different EQs is that you are creating, in certain frequency bands, phase difference between the left and right speakers or channels. And, if you're doing that in MS mode, the phase between the channels does not change that way. It only changes the phase between the mid and side. So, it's not that much of an issue, I would imagine. And we have to remember that the curves in Soma, for example, are really gentle. The maximum phase shift is fairly small anyway. Only if you do something really, really wild, like boost and cut adjacent frequencies a lot, then you can get very steep phase angles.
Justin Colletti: Gotcha. So, if you're going to do significantly different processing left/right, or between mid and side, the best kind of EQ to do it would be one like this, where you have a relatively wide filter. That's a nice thing to know. Now, we've gone really deep on some of these ideas, which is really exciting and fun to get really nerdy with you guys on this. I'd love to bring it back around to the plugin itself.
Plugin Alliance and the engineers worked really painstakingly to emulate the Soma EQ, and they worked in very close cooperation with you, Jonte, to do so. I've been told that it was one of the most advanced modeling jobs they've done yet to try to really recreate all the unique aspects of the tone of this box. I'd love to get both your perspectives on it, starting with you, Zino. You've used both of these now, the plugin and the hardware. What do you think of the plugin version of the Soma EQ compared to your hardware?
Zino Mikore: I really like it. I feel at home. Everything feels almost like there's so much me in this Soma EQ because I've been using it so much. I have this Atmos studio so that's why I love that you just came out with it because I really needed something like this, an EQ I know inside and out that can do everything.
It's really hard to compare with the hardware I've tried, but obviously I can’t use it when I'm using the hardware on the same place in the chain. So if I use it before the analog, now, it sounds a tad darker than the hardware. If I use it after the analog, it sounds a tad punchier, brighter, open.
So it's really, you can’t compare them directly because I can’t use them where they're at. It's impossible. And also with gain staging. I don't know what gain staging the plugin uses, but I use certain gain staging in my analog.
So it's impossible really to compare, but I find if I have an idea about something and I use the plugin, then I get exactly where I want to go. It's just wonderful. I've used a couple of software EQs before in mixes, back and forth. And now I only use this, which is quite funny,
Justin Colletti: Wonderful. And that's a great application for it–the Atmos mastering when you're doing multi-channel mastering,
Zino Mikorey: I think it's mixing, but for me, it's mixing, it's not mastering.
Justin Colletti: Gotcha. Oh, so you're doing Atmos mixes now, as well as your stereo mastering?
Zino Mikorey: It's like a stem mastering, but I would say it's mixing because all the panning creates so much sound. I use reverb. And, in mastering, I think nobody would use reverb or you wouldn't change the pan necessarily and balance stuff.
So it's like an extended stem mastering, but for understanding, I think it's more of a mixing. I'm looking forward to doing Atmos mastering, getting an Atmos mix and mastering it. But I think it's a little too early in the, how do you say, in the industry to do this.
Justin Colletti: Yeah, that makes a lot of sense. I have heard mastering engineers say, “I consider stem mastering basically mixing.” And when you are creating an Atmos master for people based on their stems, that is a little bit more like mixing than mastering. So I definitely understand that.
Zino, what were your thoughts of hearing this thing modeled? How closely do you feel like it compared? What are some of the things it got right and how do you think it turned out?
Zino Mikorey: Yeah, I think it's as close as I want it to be, or as I hoped it would be. It's a little faster, a little more punchy-sounding, but that's in the box, so you can't really compare it.
But, for example, if you boost 820 Hz on the third band and you boost 820 Hz on the second band, they have a different Q. The 820 Hz on the second band has a little tighter Q and there's a little bit of an air boost happening. And if you boost it on the third band, it's a wider Q, but there's no air boost.
I think the air boost is like 25 kHz, 30 kHz or something. This was the first thing I checked on the analyzer, if it does this, because I love it so much when you have the frequency and then you just go click, click. Obviously, on the hardware, you can do it at the same time.
On the plugin, I have my presets now where it's like a 1 dB and a 2 dB boost, and I just check those presets against each other. And there's just two sounds of the same frequency. So it's like two Q in one, or also the interaction between the bands seem just like I remember it.
For example, cutting the low end. I rarely ever do it because I love the sub octave so much, but if I need it, when there's an 808 or a sub-kick, something deep is overpowering, cutting, just like a low shelf, 39 Hz, 1 dB, 1 and a half dB has this quality where you have the feeling that the rest of the spectrum lifts. And it's because you're basically, correct me if I'm wrong again, Jonte, you're actually cutting here, but everything else–
Jonte Knif: You are also creating some harmony, except at the same time to the frequencies you are cutting. So that might be the thing.
Zino Mikorey: Yeah, so you get this upper bass focus.
Jonte Knif: Yes, yes.
Zino Mikorey: And it just does it. It goes, “Click, click, click” and then AB and I go like, "I'm home. Done."
Justin Colletti: Wonderful. Now, Jonte, you worked very closely with the Plugin Alliance team in designing this plugin and making sure it was as close as possible to the original hardware. How did you think it turned out? Did anything meet or exceed your expectations and what were your overall thoughts on the emulation process?
Jonte Knif: Yeah, there was quite a lot of back and forth and exchanging of information and versions. Well, the result is very good, for sure. I decided to be quite demanding because if it's not correct, then what's the reason to make it, if it's not correct?
Justin Colletti: Right. We don't want a Soma EQ GUI on some other EQ. We want Soma EQ in the computers as close as we can have it, right?
Jonte Knif: There were a couple of phases where I had to point things out. And then it took a few more months and they talked about the difficulties there were in the modeling. At some point, my understanding of DSP stops or ends, so I could not help as much as I should have maybe. But, in the first emails, I suggested a couple of things, which I think ended up being done pretty close to what I thought the correct method could be to proceed.
Justin Colletti: Well, just being able to pinpoint what the differences were between the hardware and what you are getting back from the emulation is definitely helping, even if you're not a DSP engineer to integrate the solution.
I'm curious, how did you make these determinations early on when it was still in development of what could be improved? In what way were you listening to it or analyzing it to notice what some of the shortcomings were so that they could go ahead and fix them and make it even closer to the hardware?
Jonte Knif: In the early phase, there is absolutely no need to listen. You just measure what needs to improve or lacks detail.
Justin Colletti: So you're looking at white noise through it and running each–
Jonte Knif: No, no white noise. Sine sweeps. This creates sine frequencies, different levels. And then you try to separate the output amplifier's distortions from the band's distortions. From the beginning, the interactions between the EQ curves between bands were all correct.
Then, the distortion modeling, that was difficult. And the output amplifier is sort of easy, but I think that the distortion acts up when you do some weird curves. That might be actually pretty much impossible to do in digital with reasonable processor time. In a very early phase, I told them that I know what happens here if you do weird and wild things. My hunch is that some things will be impossible, but everything else you have to do. So there are certain points where you can’t yet go.
Justin Colletti: And if you had to put a number on it, how close do you think the final emulation got to the EQ being used normally by a normal, sane person? You know, not necessarily turning every knob up to 11, and the narrowest Q setting, but when it comes to the average mastering engineer using this, how close do you think the final plugin got?
Jonte Knif: I think it's 10, yeah, because when I measure, I measure frequency response and distortion behavior with normal settings and it's correct. It's definitely correct. And there is also no sign of aliasing. It's very good. Digital technology creates some differences, like there has to be a reconstruction filter after the downsample and they put that around 35 kHz. Obviously, the hardware goes higher, but there has to be this kind of filter in the plugin.
Justin Colletti: Hopefully that doesn't make too much of a difference to most of us.
Jonte Knif: I would think not.
Zino Mikorey: Jonte, how much do your units differ from unit to unit? Is there... I mean, it's all hand-wound, etc.
Jonte Knif: We have improved the accuracy of the coil winding year by year. And we are currently within plus and minus two percent in the frequency values.
Zino Mikorey: Wow.
Jonte Knif: They are pretty much equal nowadays, but the output amplifier tubes, they are of course, individual tubes. They have a bit different distortion characteristics from unit to unit.
For Soma, I choose the tubes to create between 0.2% to 0.35% of second harmonic at typical levels. That's where it is. But you can get a bit cleaner or a bit dirtier units, depending on the tubes. The model was based, if I remember correctly, around the lower end of the distortion. I think the tubes were around 0.2% second harmonic in the mid-range for their unit.
Justin Colletti: And you do have the possibility to turn it into a dirtier one.
Jonte Knif: Yeah, that's one of the things in which the plugin is better than the hardware, obviously, because you can trim the headroom. So then you can make it really dirty or really clean.
Justin Colletti: All right. Well, I think this has been a fascinating conversation. We've gone down some deep and interesting rabbit holes here, and you guys have asked some questions of each other, which I think is really interesting as well.
Thanks so much, Jonte Knif, for joining us and Zino Mikorey. Really wonderful to have you here talking about this EQ. Thank you guys for hanging out with us.
Jonte Knif: Thank you.
Justin Colletti: See you next time.
Zino Mikorey: Thank you very much.