Inside McDSP with Colin McDowell

By in Live Sound, Music Creation

Colin McDowell is a true pioneer of the plug-in industry. Since founding McDSP in 1998, Colin and his team have created everything from authentic analog emulations to cutting-edge, “out-of-the-box” designs—most recently the new ML8000 Advanced Limiter. And not only are McDSP plug-ins the go-to tools for countless studio engineers, but an increasing number of live sound engineers are using their plug-ins on some of the world’s biggest tours. I recently sat down with Colin to learn more about the company’s history and design approach.

Engineer German Tarazona on tour with J Balvin using McDSP plug-ins and VENUE | S6L

DH: Okay. So, why don’t we just start a little bit, you know, tell me about the starting of McDSP, just the background. I know some of it, obviously, but tell me about how you came to be McDSP.

CM: Well, a long time ago I worked at a company called Digidesign—you may have heard of them—and I used to do different types of things for Pro Tools third party support, plug-ins and things like that and eventually found myself in the plug-in group, which was a ton of fun. Following that I decided to take a job at Dolby, which was all well and good, but it wasn’t as much fun as making plug-ins.

At Dolby they had a little private library that was Ray Dolby’s, and in there he had some folders about how he started the company and it sort of seemed manageable. So, I thought, “Well, wouldn’t it be a crazy idea to quit my perfectly fine day job and start my own plug-in company.” That was around 1998. I think, like a lot of folks who work in the music business, you sort of know what you like. And so if if you’re making up a plug-in, if it seems like something that I would use, hopefully other people will use it too. And therein began some of the first products, like FilterBank and CompressorBank, a multiband version of CompressorBank called MC2000, and something else called the Analog Channel plug-in, and we sort of took it from there.

It became pretty clear in the first year or so of the company that the modeling of the gear would be pretty manageable. It was either rent equipment or visit the studios, and we could make up a prototype that we could give to people that own the equipment or own the studio and say, “What do you think?” And they’d be like, “That’s pretty damn close.” We’d go, “Yeah, we thought so too.” Cool, we have a knack for this. Great. But, what we also experienced along the way was that people would say, “Oh, I wish this box had an additional mode,” or, “Wouldn’t it be cool if the gain ranges were larger, or the frequency ranges were wider.”

McDSP's brand new ML8000 Advanced Limiter

So I just sort of got into thinking that way. So, like with FilterBank, the shelving EQ’s got gain, it’s got frequency, but it also has three controls called peak, slope, and dip. It can actually control the overshoot of the shelved portion of the response, the undershoot of the non-shelved portion of the response, and the slope in between. And those are like the three basic design parameters for any shelving EQ you could ever come up with, analog or digital. I thought, “We should just put that in as some features of the plug-in.” And some of the customers appreciate it, and some of them never touch it. And it was just kind of cool at that time in the early days of plug-ins to show people, “Hey look, it can sound like this equalizer, this compressor, but then look what else it can do. Not only can it sound like some of these things you already have, but look at what additional features it has.”

Another whole aspect is live sound, where sometimes it’s about making it sound like the studio, and other times it’s about solving problems. Maybe the guitar is a little bit deader today because it got too much sweat on it, or maybe something in the sound isn’t quite right, so you have all these little nook and cranny controls like peak, slope, and dip. You can go in there and really wrestle with the sound if you choose to do so in an effort to make it sound more like what the record sounded like or how you think the client wants it to sound like.

The new VENUE systems like the S6L use Avid’s new HDX cards, and there are several products that we still have yet to release, because back in the day when it was TDM, there was either no way we could do them or it would be very, very difficult. I still curse up and down when I think about the ML4000, which is a very popular multiband dynamics limiter product that we make and is used by many live sound customers—making that work on TDM was extremely difficult. With the new HDX hardware being floating point, the way Avid has configured it, and how it’s programmed, we can now do all these things. The AE400 equalizer, which was nominated for a TEC Award last year, is a great example of, “What’s this new HDX system capable of?” The AE400 is a great example of what it can do—the AE400 would never have happened on the old TDM system.

AE400 Active EQ Plug-in

DH: So when you launched McDSP, that was really at the bleeding edge of modeling analog outboard gear, along with companies like Bomb Factory and others, right?

CM: Using an audio workstation and plug-ins is pretty much the norm these days, but you back up 15 years and I think there were still a fair amount of people that were still warming up to the idea. Not even necessarily the analog versus digital thing, but the concept that some of your signal processing could be done on the computer—not just mixing tracks together and volume.

Timewise, Bomb Factory was around the same time McDSP started. Bomb Factory went for the look, if you will, as well as the sound. We just kind of went for the sound. So, it all just kind of went hand in hand. I think there were customers that just wanted an LA-2A, and they didn’t want to have any other things to fiddle with, while other people that wanted something different.

An example of this can be found in CompressorBank, our compressor plug-in. There’s a control called “BITE”: bidirectional intelligent transient enhancement. And what that feature does is make the compressor fail, because we’d measure a bunch of older compressors like a Fairchild side by side using the same test signals and we’d say, “Oh, this is great. We can measure four of the same units. We’ve got four reference measurements.” But then we’d get halfway through, and while two units might be pretty similar, the other two units just seemed to have something wrong with them. And at some point the tech might come by after hours and say, “Oh, those two don’t work.” And we’d go, “Oh, thanks, you could’ve told us that!”

But, somewhere between working and not working there might’ve been a good sound to have. That’s kind of where the BITE control came from, so that you can set the compressor any way you want. If you want to make it fail or pass, like the higher transients almost completely uncompressed and listen to the compressor kind of trying to recover—that’s what it does. Some of the older compressors have a frequency response that goes out to 14 or 15 kHz, and when you blast your 96 kHz session through it you can hear the compressor affect sound a little differently. And for some people it’s splitting hairs, and for some people it’s like, “Wow, that’s radically different.” It depends on the kind of material you’re putting through it. I just thought that would be a cool feature to have in the plug-in, and I’ve tried to position McDSP that way. We can model the traditional stuff, but then we’ll throw on some of the extra control unit, and it seems to have worked out. You probably also noticed that our first four products all had kind of a green color theme, ’cause that’s somebody’s favorite color here (maybe it’s mine)!

DH: Tell me about the patent that you have—what product is related to that technology?

CM: The patent is regarding ML4000 and has to do with how the it acts as a limiter, and the process by which it limits signals. We were working on that back in 2006, the application was submitted around 2007. We actually received the patent recently in 2013.

Without giving away trade secrets, we describe how the limiter does its stuff, how it uses signal statistics to come up with some type of limiting action that is as transparent as you can get. There’s a knee control on the ML4000 that goes from 0 to 100 percent. Zero percent is like current modern day, loud as you can production. Twenty-five percent takes you back to the 90s. Eighty percent’s like the 80s. Seventy-five percent is like maybe the mid-70s, towards 100 percent, “congratulations you’re back into the 60s or the early 70s”—the heyday of dynamic music production.

But it does that because it doesn’t operate like a normal limiter, so we thought we’d submit a patent application, “lawyer-up” as they say, and have it done all properly. And in the meantime all these loudness specifications have come out. The ML4000 was an interesting product in that regard. Something that we thought was unique and cool.

We’ve licensed ML4000 patented technology to gaming companies like Bioware and Ubisoft, and worked it into LouderLogic, our audio-player iPhone app. The New York times called LouderLogic “perhaps the most impressive audio enhancement app [they] have used”, so I suppose its good to have that algorithm patented!

FOH engineer Snake Newton used the ML4000 on the latest Duran Duran tour with S6L

DH: When you port a plug-in for a new platform, do you try to make it sound exactly like it was on the original platform, or how do you approach that?

CM: That’s a fabulous question because a lot of customers ask us that. I can’t cook, I can barely ride a bike, I can kind of ice skate, but boy can I model old stinky analog equipment that’s used to process audio, and I’m really good at making floating point C code behave exactly like fixed point assembly code. When we originally got to making plug-ins, there was that challenge of making the native version sound like the TDM version. I’d done it at Dolby for a while, and that was something I had a knack for. Probably the biggest takeaway I have from working at Dolby is that I would be given something by the R&D group, the research guys, and be asked to make it into a product. All their work would be in C code or MATLAB, or what have you, and eventually the product people like me would have to make it work on something like a Motorola fixed point DSP.

So when it came to the HDX platform, we had an HDX Native solution already, so making the new floating point HDX DSP version was pretty doable. But the reason your question is so interesting is that with the new chip you have a new paradigm, a new way of doing this. With floating point, now the signals can go above zero dB. Internally, the system itself has a lot more dynamic range. This is especially useful in live sound where stuff just happens that results in a significant level change. You have to sort of roll with it. Whereas in the older systems there’s a danger of potentially clipping, now that sort of element is removed—one less issue to worry about.
Now you have all these algorithms, this body of work that’s pretty much assuming everything we get is going to be below zero dB. All of a sudden you have a new platform with way more dynamic range and fidelity, and it’s like, “Oh, wait. What can happen?” You have to have some additional smarts to make your algorithms operate in such a way that they could handle a signal that was below zero dB or well above it—that was kind of a cool challenge. In hindsight I can say that at the time we were like, “Holy crap, what’s Avid doing making this stuff so much better? It’s giving me a headache.” Anyway, so that’s the challenge about the HDX hardware that’s really cool.

The McDSP team

DH: How does McDSP work with the VENUE team to qualify the suite of plug-ins for S6L?

CM: That’s a good question. Part of it is a little bit smoothed over for us because we already make venue-compatible installers. So, our whole builds process turns out venue-compatible installers already. In fact, the ones that we just churned out for version 6.1 includes venue-compatible installers for the very latest plug-in we just made, like the SA-2 dialog processor, which probably will have some good uses in live sound. And out of the gate, our new plug-ins have S6L page table support, the control knob layouts for each plug-in.

VENUE | S3L-X system at McDSP used for developing and testing plug-ins

DH: In addition to your emulation plug-ins, you have other plug-ins like FutzBox—how do those come about?

CM: Sometimes we make products, and they have come from, “We think this would be pretty cool, and we hope our customers will too.” Sometimes customers will take products we’ve already made and say, “Hey, this is really cool, could you make something else?” FutzBox was pretty much a customer saying, “All your plug-ins sound good. You need to make a plug-in that sounds bad.”

“I beg your pardon?”

“You know, screw up the audio, distort it, filter it, make it sound like it’s going through a walkie-talkie.” And after you talk to enough post customers, you realize that the notion of doing FutzBox or futzing is a bread and butter everyday kind of thing. We should make it a plug-in, and so we did.

I think another good example is our new SA-2 dialog processor. Some people might say, “If that’s just kind of like an active equalizer, why can’t we just use the AE400?” The answer is that SA-2 is modeled after a different box made by a post-production and re-recording mixer named Mike Minkler who came to us and said, “I see what you’ve done with this thing, the AE400, and I have something that’s kind of like that but not really. Could you come take a look at it?” So we took a look at it. “Yeah. It’s kind of like that, but not really.” “Yeah. Could you make one?” “Maybe.” Make a prototype and back and forth, and he’s really happy with it. Okay, cool, make it a plug-in.

Mike Minkler's one-of-a-kind processor on which the McDSP SA-2 Dialog Processor is based

DH: Finally, among other things, McDSP plug-ins are known for their extremely low latency—what can you tell me more about this aspect of your products?

CM: When McDSP started we were still writing plug-ins for TDM, and with TDM you’re down there at the assembly code level—you can almost look at it, like a window of opportunity you have to get your inputs in, and push out your outputs. It’s almost kind of like a fun game, and in any case it seemed best to have either no latency or as little latency as possible because that’s why people were buying TDM hardware. Even if you fast forward to today, no matter what the system, people still seem to be all about their latency. It just so happened that we got started when TDM was the thing, and I actually like writing assembly code. And in that type of approach for writing algorithm code, be it audio or any other type of signal, the notion of having a non-latent algorithm is just kind of inherent.

If you want to record more tracks, you’ve got to put the headphones on. And when you hit play and you start singing or playing or whatever, if there’s some kind of crazy weird slappy delay echo thing going on in there, it’ll drive you crazy. That still seems to be an important feature to customers.

I think it’s a happy accident, but the live sound market appreciates—perhaps even more than the studio people—the latency in an algorithm. For the live sound community, it’s not just a convenience issue. If the system is at front of house and going to distribute the audio to the masses, the system has to be as low latent as possible to deal with the other latency issues you have to deal with, like aligning the speaker array, etc. If you’re mixing monitors and the in-ear mix is six milliseconds back, sorry, your clients aren’t going to put up with that.

A collection of some of McDSP's industry awards


VENUE | S6L Now Available

The next stage in live sound is here—with the award-winning VENUE | S6L system, you can take on the world’s most demanding productions with ease.

Sr. Marketing Manager for Avid Live Sound Systems and Music Notation. I previously worked at Euphonix and E-MU Systems before joining Avid.