Issue link: http://digital.copcomm.com/i/987065
C A S Q U A R T E R L Y S P R I N G 2 0 1 8 57 MEET THE WINNERS full-time professional applications and have somebody use no-compromise audio inputs, analog limiters and all those tools that professionals expect, in a product that has the right compromises for them. How did you decide to do a 3, a 6, and a 10? From the earliest days, small was always something that was important to us. The 744T represented the next thing after DAT, and we came with the 744 and the goal was let's have as small a product as we can that has all the capabilities of DAT plus a couple of extra channels in this file-based world. We've always been attracted to keeping things small because we know many of our customers are working out of a bag, and we're really conscious of that. We also see the world of cameras and the visual side getting smaller and smaller. That, to us, is important. For us to have products that are consistent, lightweight, capable. When we said, "Let's come up with a product line," we wanted something that was a boom and a couple of wireless. That was the MixPre-3. Then something that was the classic four-input. That was the MixPre-6. And the MixPre-10 is the classic eight full microphone input device to bring to a new customer. But your existing customers recognized the MixPre's right away, didn't they? We've also seen that existing customers have gone down this path as well because this is something they want to add. The things that are exciting are that we're able to put timecode and eight Sound Devices great mic pre's in a super-teeny product. Again, there are compromises that, for its target customer, may or may not be an issue. Also, what it represents is a capability for us to engineer and build a product with so few of those compromises at that price point. It also represents a tell of what we're capable of doing in future generations of our full professional products. It seems natural that once the Mix-3 was done, it was not a stretch to build a 4-input and 10-input model. Not minimizing hardware design, as it's something so core to our ethos, and that has come out of Matt Anderson from day one. But what we've also seen is that it's a world of software. The first microcontroller-based product we did was the 302, then the USB Pre and the 744T. That was a super-challenging product and, frankly, some of the earliest versions of 744T firmware were really pretty rough, but we were absolutely committed to making these products as robust as they could be. When you look at where the MixPre is, from a software evolution standpoint, we are building products that are so far more capable and robust compared to the professional products from 2004. You see the evolution of the company, you also see an evolution in the industry and the computational power that's available for manufacturing. That's the exciting thing. It used to be that professional products led the consumer world and now, in many ways, the consumer world is where some of the highest technology ends up first because it has to be made at such scale, and we're able to take advantage of that. Outstanding Product – Post Production iZotope RX 6 Advanced by Bob Bronow CAS My name is Bob Bronow CAS and rough audio has been part of my life for a very long time. I mix Deadliest Catch. As an early adopter (RX 1), I've seen RX become an indispensable tool to our industry. Since then, RX has grown in leaps and bounds. iZotope won its first CAS Technical Achievement Award for RX 3. Since then, every version of RX has won the Cinema Audio Society Outstanding Product Award. When I wrote the article for RX 3's win, I was mixing Season 9 of Deadliest Catch. Five years later, I'm mixing Season 14 and RX is more valuable than ever. Every new version of RX makes "real" things that were previously impossible. New modules in RX 6 Advanced include Dialogue Isolate, De-Rustle, De-Wind, Spectral De-ess, Mouth De-Click, De-Bleed, and Breath Control. What seems to be the big jump between RX 5 and RX 6 is the introduction of "Machine Learning." (To explore more on Machine Learning/ Neural Networks, check out CAS member Jay Rose's article on "Neural Networks" in our Winter 2018 issue.) I caught up with Alexey Lukin, Principal DSP Engineer, and Mike Rozett, RX Senior Product Manager at iZotope, to talk about the new technologies incorporated in RX 6 and where things might be heading in the future. CAS: RX 6 is the first time Machine Learning has been incorporated into RX. It allows RX to do things it couldn't do with previous technology. How did you come to the realization that you needed a new technology to deal with these problems? Alexey Lukin: The problems are getting harder with each version of RX. A lot of the conventional algorithms that we used in RX prior to RX 6 have been based on the same kind of approach: The programmer has to explain to RX exactly what to look for and how to identify the problem on the spectrogram or in the waveform. Once the problem has been identified, the programmer writes an algorithm, step-by-step, for removing the problem. This method worked well for five versions of RX. But it became very difficult to explain certain problems, like