Issue link: http://digital.copcomm.com/i/1259912
68 S P R I N G 2 0 2 0 C A S Q U A R T E R L Y re-recording mixers but also for dialogue editors. We heard that people in the business were having a lot of issues with trying to cut [smoothly] between different dialogue takes. You get a lavaliere take, you get a boom take, and maybe you have ADR and, in some situations, you are cutting between all three of those to come up with your ideal performance. There's a lot of pain generated from and time spent on trying to get ADR or a lav to sound like a boom. So, the idea was: How can we save time and automate the process so that you can cut between those different mics and those different recording situations? It came from a lot of user interviews and people talking about how they are spending a lot of time with EQ and reverbs and ambience and how it would be great if [iZotope] could help with that. And that's where it really all came from. You just talked about the three main parts that comprise Dialogue Match. Do you feel that covers everything you might need for the task? We had a lot of discussions when we were mapping out this product about what are the essential elements to learning what a sound is and then being able to apply it to another track. What we settled on was that EQ, reverb, and ambience were three of the most essential elements that go into learning the sound of dialogue and applying it to another one. There are lots of other variables that you can play with and one of the requests we've been getting from people is pitch, "Could pitch be part of dialogue matching?" The short answer is absolutely. There could be things like [pitch] that come down the road but, for now, we feel really good about this set as we go toward the next version of the product. Had iZotope been working on reverb independently or did that come along with the acquisition of Exponential Audio? A little bit of both. iZotope had done some work on reverb and it was in a few of our products, in fact, it's still in one or two. Dialogue Match was the first thing we've released that combines Exponential Audio tech developed by Michael Carnes with brand-new and existing iZotope tech. We were thinking about this as a fusion product that would leverage Carnes' work with our own machine learning and DSP. It was cool to offer the first version of its kind, released after Exponential Audio joined us. You mention machine learning. Do you see machine learning as being something that will integrate into most of your products moving forward? I would say that machine learning is getting to be a bigger and bigger part of what we're doing. It's something that the team got into relatively early in the audio business and it's one of the things we do incredibly well. I'm not sure that it's going to make it into every product that we have, but it's definitely going to grow and make it into more and more of our products. It has allowed us to solve problems in a way that you couldn't do before. Artificial intelligence has helped a lot of industries innovate, and it's no exception in audio. Machine learning allows you to do some incredible things like separate a source. There were some EQ tricks you could do to separate different bands and side processing—and they were pretty effective at the time. But with machine learning, you can take a file and break things into parts and stems. That's still really early in its implementation and I think going down the road with it, I think we're going to do some things that even a few years ago seemed impossible. Was there anything unique about the development process for Dialogue Match? We had to develop the reverb matching process from scratch because no one had done that before. Then we brought in reverb technology from Exponential Audio and combined it with iZotope's EQ and Ambience Matching tech. We had to get those four things to work together and then add intelligence that learns the sonic profile of the reference and target tracks. That kind of collaboration and development was very unique and something our team did a great job on. That sounds pretty intense. What would you say has Dialogue Match been used for that I might not have expected? One of the things we thought it might be used for is with localization. We knew people would use it entirely within one language, matching up boom, lav, and ADR, but we had some ideas about taking the profile of a particular language and applying it to another. We let our beta testers inform us without leading them and some people immediately gravitated toward this. It was cool because we thought people would pick up on it and they did. the things we do incredibly well. I'm not sure that it's the things we do incredibly well. I'm not sure that it's the things we do incredibly well. I'm not sure that it's iZotope's Dialogue Match showing the three main adjustment parameters: EQ, Reverb, and Ambience.