The name Dugan has long been known as the authority in automatic mixing, a technology that intelligently turns microphones on and off. Dan Dugan, the man behind the famous Dugan Speech System, recently partnered with the audio experts at Sound Devices to implement his automixing technology into the Sound Devices 688 mixer/recorder. This is the first time a field production mixer has included the Dugan system, and that’s a pretty big deal. Even though automixing has been around for nearly half a century, many location mixers are unfamiliar with it. So we talked with Dan Dugan and Jon Tatooles, Sound Devices’ co-founder and chief business development officer, to find out more about this collaboration and what it means for the pros in the field.
Dugan explained to us the basic need that automixing fulfills. In any situation where you have multiple microphones and several people speaking within the same environment, you’ll run into some inherent challenges. “The main problem is that you have more microphones picking up the noise in the room,” says Dugan. “The more mics you have, the noisier the sound becomes. Even with only two microphones, you’ll require active mixing, which is when the mic on whoever is not talking needs to be turned down and the mic on whoever is talking needs to be turned up. And when someone else is talking, their mic needs to be turned up and the other mics need to be turned down.” The more mics there are, the more important this active mixing becomes. And the tougher the job becomes. “A lively discussion panel can be very difficult to follow, especially when the sound operator isn’t sitting directly in front of the people talking and can’t see things very well, which is very common. So it’s not unusual to have that up-cut, which is when the sound comes on late. It’s very common to hear that in sound for live broadcast.”
This same situation occurs in sound reinforcement with regard to feedback. Dugan continues, “You may have plenty of gain with one microphone, but if you try to have a panel where six microphones are on, there’s going to be feedback. So noise and feedback are two problems that happen with multiple mics. A third problem can occur, as well, known as comb filtering, or phasing. This happens when you have two mics that are fairly close together – for example, two actors facing each other with both actors wearing lavalier mics. One person’s voice goes into their own mic, but it also goes into the other mic with a slight time delay. That causes a frequency-response coloration. This problem is also solved by fading up the taker’s microphone and fading down unused microphones. Unless it is completely scripted and rehearsed, this can be very difficult to pull off. So there are a few sub-problems, but the overall problem is this constant need to fade up and fade down.”
Dugan first encountered this issue in 1968 while working on the musical Hair in Chicago. “In that production of Hair,” says Dugan, “which was in the days before having wireless on everybody, there were, if I remember correctly, sixteen hanging and footlight mics, ten mics on the band, nine hand mics – three stage right, three up stage, and three stage left – and one wireless mic used for a special effect. And this was back when you couldn’t buy a mixing console. Back then, mixing consoles were hand built by the chief engineer of a radio station or a recording studio. We would mix on a rack of rotary-knob mixers. You can imagine trying to handle that number of microphones on a rack of rotary-knob mixers. It was pretty much impossible.”
It was from this experience that Dugan came up with the concept of automixing. “I experimented for about six years, just trying various different things and breadboarding circuits and stuff like that,” Dugan continues. “I came up with a working solution that uses a microphone to monitor the ambient sound level in the room and then gates on microphones that come above that ambient sound level.” Dugan then further refined the system to include a downward expander instead of a gate, providing smoother action. “The adaptive downward expander on the inputs was complemented by a number-of-open-microphone (NOM) attenuator on the mix bus to prevent noise buildup or feedback. That system is what’s now called the Dugan Music System, which is available in some Dugan products.”
“I patented the Music System and demonstrated it at the AES convention,” says Dugan. “Then, before I was able to make any kind of a deal for commercializing it, I discovered a new principle while continuing my experiments. I had been thinking, ‘When there’s one person talking and all the other mics are just listening to the ambience, there should be some way to measure the ambience out of that. You have all these other mics listening to ambience, so how can we do this somehow without having to have a separate mic for the ambience?’ I was just fiddling around, really, and experimenting and trying different combinations, and I tried an arrangement by which the sum of all the inputs before the processing was the reference. And all of a sudden, something magical happened. This function happened that when someone talked, their mic came up and the other mics stayed down. And the gain stayed constant. It was amazing. I can’t say that I invented it any more that I discovered it just through tinkering. I applied for a patent for that principle, which I call the Dugan Speech System, and that’s the system which is commonly known as Dugan today.”
But Dugan didn’t know exactly what he had done. “It just kind of fell out,” Dugan explains. “In order to write the patent, I had to analyze it and write the formula for it, which turned out to be very simple. It’s what a mathematician would call elegant: The gain of each channel is calculated to be the same as the ratio of the level of that channel to the level of the sum of all the channels. I’d be proud if I had written that algorithm, but I didn’t, actually. I had to analyze that out of what worked.”
Revealing the secret recipe doesn’t bother Dugan, as his system includes refinements that cannot be duplicated so easily. “I added some frequency pre-emphasis in the side chain,” continues Dugan. “The time constants are very important, as well. In fact, that patent is expired, so there’s a generic name for that system, which is called Gain Sharing. So I’m the inventor of Gain Sharing, which is the generic term for the Dugan Speech System.” A number of what Dugan calls his “honorable imitators” have manufactured products that accomplish similar tasks. “They do a gain-sharing system that’s not exactly the same way I do it, but it’s the same general principle. And they don’t use my name, of course. My system has the parameters tweaked, and when you license a Dugan system, you know that you’re getting exactly the right thing. So that’s how I’m still selling an expired patent, because of the expertise in doing it the best way.”
Before Sound Devices brought the Dugan Speech System to their 688 mixer/recorder, the 688 already had a different automixing function called MixAssist, which the 688 still includes. Jon Tatooles of Sound Devices explains the details. “The recent firmware update adds Dugan along with MixAssist, so now the customer can choose between the two algorithms,” says Tatooles. “At the end of the day, they both accomplish the same basic goal, which is to improve the mix by activating used microphones and turning off unused microphones. But the way they do this is different. The Dugan system is very well-respected and well-regarded for live-broadcast applications in particular. So that’s one of its core strengths. We introduced the MixAssist algorithm in our 788T recorder, and it’s a great algorithm. It may be better suited for tougher applications where you need it to be a little more aggressive. The MixAssist algorithm attenuates microphones in a different way than the Dugan algorithm. Whereas Dan’s algorithm is super smooth, MixAssist may have more noticeable transitions. But in some cases that may be the right tradeoff. I think the user is the best person to choose which one is the most appropriate for their application. They’ll get a sense of it in different environments by switching between the two algorithms and finding which gives them the sound they want. The nice thing about having that tool within the 688 is not only can you generate that mix with MixAssist and Dugan, and output that – let’s say you may output that to camera – but you can have all of your pre-fade ISOs on individual unprocessed tracks, so it can be remixed later. You have lots of flexibility.”
Dugan tells us that he did not compromise his system’s performance when integrating it into the 688. “It’s the full-bore Dugan Speech System performance,” says Dugan. “I’ve tested the basic function very carefully, and it absolutely performs the guaranteed Dugan automixing function. It is stripped down in terms of controls, compared with a standalone Dugan. For example, it doesn’t have separate on-automatic-off, and it doesn’t have a separate weight control for each channel. But in most situations, most people don’t use the controls very much. I rent out automatic mixers every week to people. They’ll often come back and nobody’s touched the controls. It works fine right out of the box.”
On this topic, Tatooles says, “We have made Dugan very simple for the 688. We’ve consciously left out some of the tools that are available in the standalone Dugan products so that it’s a simpler implementation. We want people to be successful with this right away.”
As automixing technology is relatively new to production sound, many production mixers may not be fully familiar with how to properly utilize automixing – when to use it and when not to use it. Dugan explains how to approach this in the field. “Well, it’s called the Dugan Speech System for a reason, and that is because it’s optimized for speech, which does include rap,” says Dugan. “But you don’t want to use it to mix a musical group, though it could be a special effect. For example, you might use it on some elements of a drum set as a special effect. But it’s definitely not intended for people harmonizing with each other, because the algorithm separates the voices from one another, which is the opposite of what you want to do in music. In music you want to bring them together to blend.” Dugan goes on to say that in situations when people are going to be singing together, like in musical theater, users should switch out of the Dugan Speech System into manual mixing, or switch to the Dugan Music System if they have that feature.
“So the speech system is intended for speech,” says Dugan. “But beyond the caveat that you shouldn’t try to mix people singing together with it, it’s really universal in terms of application in any place where you’ve got a bunch of people talking. The Dugan Speech System algorithm can be helpful in TV news panels, sports discussion panels – ESPN uses it for all their panels – conferences where they have a panel discussion, political debates. They used Dugan for most of the Republican presidential debates, including that famous one with seven different people going back and forth, talking over each other. That used Dugan very successfully. Dugan works great for situations where you have a number of people talking: late-night talk shows, game shows, civic applications, sound reinforcement for boardrooms, all kinds of civic boards and panels, boards of supervisors, school boards, and things like that. Also, it should be very suitable for reality TV, but I don’t think that it’s much used there yet. The Letterman Show was an early adopter, and they used it for about 15 years. It’s great for dialogue microphones in theater, where you have multiple wireless microphones. And collaboration is a big one, too. A long time ago, in the analog Dugan days, I put in a system for American Airlines. They had a big conference room with 25 mics in it. It was on 24 hours a day. Any number of people could just walk in and sit down and have a teleconference with somebody. Houses of worship, as well, can benefit from Dugan, especially those that are more speech-oriented than music-oriented. It benefits many Catholic churches, where you have people who talk from two different podiums, and you might have two or three different locations on the altar where you want to pick up speech. The Speech System works really well for that. So the applications are basically numberless. I’m working on making products that can be applied, and there are lots of areas that have yet to be tapped.”
Tatooles elaborates, “Production automixing is best suited for unscripted productions where you have multiple people on set, and at any moment, one of them is going to jump in. Also, these algorithms have the ability to turn off the automatic mixing tool on any given number of microphones. For instance, if you have a four-way interview application where you never know who’s going to jump in, but there’s one core person. Maybe that’s a head of state or something, and you want them to always be on. So you can turn it off on that mic.”
“Some applications are different,” says Tatooles. “If for example, you’re in an incredibly high-noise environment and you have one person close to a noise source and another person that’s far from the noise source, you will hear a change in background noise when the transition happens from one mic to another. So at that point, you have to decide, if they’re far enough away from each other that they’re not interacting and multiple microphones are not picking up the same sound source, then an automatic mixer is going to give you less of a benefit. And because these algorithms are speech-specific and not designed for music-oriented signals, you’re not going to use this when you have speech mixed with music production or just a full music production.
Tatooles tells us that Sound Devices has plans to help educate their user base on the benefits of automixing for production. “Some people believe that they can just turn up the faders and let the automatic mixer go, and it’s going to manage everything. Well, it doesn’t,” says Tatooles. “It’s going to manage activating and deactivating mics when someone talks and someone else doesn’t. You want to reduce the number of open microphones to improve clarity and speech intelligibility and so you don’t have a build-up of ambient noise and reverberation. You’re going to do that whether you have an automatic mixer or not. The relative levels and the overall mix is always going to be the responsibility of the production mixer. The automatic mixer assists in turning microphones on and off, and it makes some relative gain changes based on how many mics are open.” Tatooles says that whether users should use the algorithms or not will be situational. “Sometimes it’s not appropriate to use automixing because there won’t be much of a benefit,” explains Tatooles. “Other times, it’s going to be the only way you’ll get a usable recording. Then there’s everything in between. This requires someone who’s knowledgeable about the strengths and the limitations of the system and who knows when it is or isn’t appropriate to use.”
We asked Dugan about the differences in his speech system now compared to its first iterations. “The algorithm is the same,” says Dugan. “It’s been tweaked through listening. I adjusted the various different variables forty years ago when my system was still licensed to Altec. And those variables are correct. So the algorithm hasn’t changed. What has changed is the realization of it. It’s done by DSP now, which means it’s much more precise and a little faster on the uptake. It was fast enough in analog with a 15-millisecond attack time. Now in digital, we’re using a 5-millisecond attack time. And now we have ones that are unbalanced analog, we have boxes that are balanced analog, AES digital, ADAT in and out, and newer products that have been introduced recently – 64 channels of MADI and another one with 64 channels of Dante. So I’m trying to fill in different connection schemes for different applications.”
“Dugan works better and is easier to use when it’s integrated into a mixer,” Dugan tells us. “My own products are insert devices that you patch into a mixing console. And then my licensees, like Yamaha, for example, have Dugan built into their QL and CL series consoles. It’s the same with the Sound Devices 688. The Dugan system is built in, and you just turn it on and choose which channels you want it applied to. It goes to the LR buses. It has up to 12 inputs, and the automatic mixing action is to the left-right buses. That’s where it’s placed in the system. Every mic has its own track in the Sound Devices, so you’ve got ISO tracks for everything. But then you also have the automatic mix on the left and right, so at the least, you have something you can turn in for dailies on a production. You have an instant mix that’s not full of the noise of twelve microphones. And that mix might be a final mix. You might not have to go back to those ISO tracks, and you can just use that for either instant on-the-air production or webcasting and things like that. You can use the automatic mix instantly on the air. Or if you’re doing post-production, you may find that the automatic mix will work for 95 or 99% of the program and you just might have to go back and use the multitrack channels to mix a few things where people are overlapped or something, and you want to make that different.”
Tatooles tells us how the partnership between Sound Devices and Dugan came to be. “I’ve known Dan personally for many years, and so has my business partner, Matt Anderson, who’s our president. Matt and I worked at Shure, Inc. I was there for about ten years, and I managed their mixer and signal processing products. Within that group, I had a few products that were designed specifically for sound reinforcement. The first one was called the FP410, with the popular SCM810 coming a few years later. The FP410 was the first portable, battery-powered automatic mixer. This was in 1991. Fast forward many years, and now we have extremely capable portable field mixers and recorders. We recognize that these mixers and recorders are being used for production and that the material is being recorded and then post-produced. But there’s still validity to the automatic mixing context for production. For example, in this day and age, there is so much unscripted dialogue and lots of actors moving in and out of scenes. Sometimes there’s simply no way to manage all of the cues. You don’t want to up-cut anybody, and you don’t want to have to always rely on going back to your ISOs to remix things. If you can get a great mix, then go for it and use whatever tool you can. So that’s why we implemented MixAssist initially. We had always known about Dan and held his algorithm in high regard, as do the many users who use it. Because the Dugan algorithm is now available in add-on cards for bigger mixing consoles and integrated into many of the Yamaha consoles, there’s this whole group of professionals who have been exposed to the benefits of the Dugan algorithm, and we wanted to have the actual Dugan algorithm. We didn’t want a substitute. We didn’t want to have a ‘Dugan-like’ thing. We wanted Dan to contribute to making this a great implementation. So we worked directly with Dan to implement that in the 688.”
Dugan adds: “I’m pleased to work with Sound Devices because I’ve known them and respected them for a long time. I’ve always admired their products, and I use them myself. Besides the invention business, another thing I do is nature recording for the Nature Sounds Society, and I use Sound Devices out in Yosemite National Park to record nature in surround sound.”
Photos courtesy of Sound Devices & Dan Dugan