5th May 2019
Many music products are conceived by experts in a particular part of the recording process, but thankfully some aren't. For example, back in the 80s at the beginning of software MIDI sequencer development, MOTU saw the computer as a solution for linear recording with MIDI and adopted a tape transport paradigm as their GUI. On the other hand, Opcode's founder Dave Oppenheim, saw the computer QWERTY keyboard as a series of buttons that could be used to trigger sequences, and merge them into larger sequences, which has led to the "stutter" functions available in more modern applications like Ableton Live. Offering a familiar paradigm makes it easy for users to get up to speed on an application quickly. But being unencumbered by legacy can lead to new ways to work.
iZotope is most known for their mastering and audio repair applications, as well as a series of interesting instruments. They didn't start out to create something using existing paradigms. In this regard their lack of experience in the recording industry has been a benefit. Although not covered in the interview, they have also created some innovative instrument plugins like Breaktweaker, Stutter Edit, and Iris.
iZotope's founder Mark Ethier follows the second philosophy of design. What you don't know about a workflow can be just as useful as knowing everything. Balance is the key. Mark, who is quite learned and easy to talk to, keeps his eye on the needs of his customers and the value that technology can bring to them.
Tell me about your background as it relates to music and technology?
My family had a lot of technology around the house. My dad had a computer startup back in the early '80s, so I got the exposure to computers and music through my entire family. When I went to college, I went to MIT (Massachusetts Institute of Technology), and studied both music and computer science, totally separate. So, I have a degree in music, very traditional, conservative background, and then I have a degree in computer science, taking on more of the applied math side.
Do you play an instrument?
Piano is my main instrument, but I actually studied composition. I thought I'd be a film music composer. It wasn't until my sophomore or junior year in college that I did an internship at Cakewalk, because they were right off the MIT campus in Cambridge.
That would have been when Greg Hendershott was there...
Yea, this was early days. Cakewalk was about to release Pro Audio 9 when I was there, and a product called Pyro, which was an MP3, CD burning, music organizing product. I realized I could combine this passion for music and this passion for technology, because before that, the two were very independent. I had synthesizers growing up, so I had some experience there, but really, I had never crossed the two.
I guess what you're saying is the most important thing you learned at that time was from the internship at Cakewalk. Were there things you learned from your education at MIT that crossed over to the music part or vice versa? What was the most helpful thing about college to you in your career?
The culture at MIT and my experience there influence the culture and the way our products are built. Unapologetically I played a lot of video games in college. When we launched our first version of Ozone, that was one of the first things people said, "This is really cool, but a little bit annoying." Because we used to have everything fly in or fade in, so it was like a video game in some ways. At MIT, there was a lot of overlap of artistic pursuit and the technical pursuit.
What was your favorite video game?
I have to admit that I played a lot of StarCraft in college. It's a realtime strategy game which combines the cerebral aspects of chess with the adrenaline, speed and team dynamics of traditional sports. And it was a lot of fun.
With video games, there's a lot of real time programming, optimized to get things running really fast, and the audio and the animations and everything that happens. How much real-time is in your products?
When we started iZotope, it was 2001, so computers were fast enough that you could do a huge amount of processing natively. So, we didn't actually need to have external hardware, which is good, because we didn't have money and couldn't afford to buy it. But it also meant that computers were fast enough that we could do a lot of really rich visualization, and that a lot of that it could happen in real time. In the early days we took the philosophy that computers were getting faster and faster. So instead of worrying about optimizing for what they would be like in 2001, we figured okay, in 2002 they are going to be twice as fast, so let's always push the boundary of what we can do.
And memory and storage would be cheaper...
Exactly, for us it was memory. If you think about trying to make algorithms or audio DSP that runs on an embedded piece of hardware, often one of the big limitations is memory. And with a computer, we had plenty of memory, so we were able to do types of processing that just was not possible. Spectral processing especially wasn't possible in an embedded environment.
How did iZotope get started?
iZotope started really as a fun project. With some of these friends, we got together and made a product called Vinyl, which made your music sound like a vinyl record. We were still undergrads. We put it up on a website, and thousands of people downloaded it. It was enough for us to feel like after we graduated, we could try to sell something. And that's when we made our first version of Ozone, and launched it at the end of the summer, partially because we needed money for food and rent.
In 2001 computers had become fast enough and musicians were in a position where they could do really great recording themselves, and things like Napster were there, and MP3s. You could then distribute your music out in the world. When we released Ozone, we had never been in a recording studio, nor a mastering studio, or used a piece of analog gear, and so we represented this new breed of home musician. We were making products for ourselves because we were trying to figure out - how do you do mastering? And in that process, we built a product that made sense to us.
When did you decide you could quit your day job and do just that?
One luxury we had starting the company when we were just in college was we had enough people who had actually downloaded the Vinyl product, that we felt that even if not a lot of them would buy something that we made, we probably could make enough money to afford rent and food which was all we needed. So, in some ways, we didn't have to quit our day job because we never got jobs. And we didn't really think a lot about, "What are we going to do six months from now?" because we were 20 years old. Probably if I did it now, I'd be much more worried.
How would you compare the impact of your computer science background, and your musical background? And which is ultimately more important?
I never had a recording studio. I never recorded before we started iZotope. And that was actually part of why iZotope started. One example of why this happened, and I love telling this story because it's incredibly embarrassing. The MIT Symphony Orchestra was performing one of my pieces, so I thought I should record it. But I'd never recorded before. I had a $50 budget and I rented some equipment from a local music store. They gave me two SM57 mics, a mixer and a DAT recorder. I dutifully set those up in this great hall and needless to say the recording was terrible. And that actually led me down the path of trying to understand what did I do wrong? Why doesn't this sound good? You can imagine now where eventually mastering, audio repair, these things start to come out, because that was my experience of trying to figure this out.
A classic case of necessity being the mother of invention. So, a lot of the products you guys make are designed to help people who are coming in to this part of recording without experience...
Yea, that's definitely infused our mindset and the culture of the company—trying to help people be creative and get rid of technical barriers. I was in that unique position where I was a musician. I had a great ear. I had been studying at MIT as almost like a conservatory background, but I also had the technical background, so I could look at the products and say, "I know the mathematics behind this, but as a musician, this isn't how my brain works." And so, we were able to bridge that gap that create products that hid the technology and made it as easy as possible for people to get a musical sound without necessarily knowing the underlying stuff.
To be honest, I didn't know that MIT had a music department. What was the typical profile of somebody in that department?
There's lots of interaction with other institutions. MIT has the great benefit of being in the Boston area. My brother was a grad student at Boston Conservatory (since merged with Berklee College), which is just across the river, and when I was an undergrad, we had some of the same professors because they tend to work across all of the different universities across Boston. So, you've got Berklee, Boston University, Emerson, on and on and on... and they'll also work at MIT or Harvard as well.
Other than the bad recording of your composition can you remember another failure that contributed to your success with iZotope?
I would say that the history of iZotope is defined by failures. When I talk to entrepreneurs that are just getting started, I think the most important thing is resilience, because the only thing I can promise you is that you will fail. And you have to pick yourself up, learn from it, and try again.
Where to start... iZotope actually had a line of photo products we made, that you probably don't know about. In the early 2000s, we had a line of photo products that were $29, one click, sharing to the web, seemed like a great idea, and then Apple came out with iPhoto. Then we had this idea that we were going to make the ultimate DVD audio authoring tool. And we all know how DVD audio left. So, there're many things that we have tried that have not worked, and really, it's because we got ourselves back up and tried again that we've continued to be able to grow.
Sometimes you need to know when to stop...
Yea, that's one of the hardest things.
So, when you entered the market there were already established products and workflows. What do you think are the critical elements to getting people that are busy and on deadlines to try a different way of working?
That's a good question. I think one of the most important things we try to practice is compassion for what people are going through. We work with post professionals who might have 24 hours to mix a TV show and have no opportunity to make a mistake in that. Every moment is precious. So, for us, some of it is honestly patience—find people, and get to know them. And so much of the professional world is especially through word of mouth. So, you really spend the time and invest it with individuals. They then get to understand your products and your tools, and then they can help to go out and explain, "Hey, here's a different way to do it." But it's really very one on one.
Do you have a process for developing products where that interaction with your customers is now a part of it?
As I said before, one of the lucky, unique aspects of starting iZotope was that we had never been in a recording studio, never recorded an album, and so part of it for us was we had to ask people to understand what we needed to do and learn from them. And that has really defined the culture at iZotope, where we are constantly interviewing people. We have dozens of one on one customer interviews that happen every week with people around the world. Either they come in person or we go to them over Skype. We have a huge amount of interaction that happens with customers non-stop.
A lot music these days is monitored on high performance gear, but actually listened to with ear buds. Ozone has a mastering reference feature, how does that element play into the different ways that people listen to music now?
I think of the important aspects of mastering is trying to take what an artist intended the sound to be, and it sounds great in the studio because you have fantastic equipment and the acoustics are perfect, and then figure out how to reproduce that when someone listens to it on ear buds, a TV, in a car, on a laptop. I think that one of the arts of mastering is figuring out how do you take something that sounds great in one place and then program into it in a way something that will allow it to sound good everywhere. Ozone has some features that allow you to preview different types of encodings and you can start to imagine what will this sound like in those environments. So that's one of the ways we try to help people bridge that gap.
iZotope makes extensive use of Machine learning in your products. How are aesthetic and subjective decisions incorporated that into the outcome?
Machine learning is interesting because I think there are a lot of misnomers about what it's good at, what it can do, what it's capable of, and I think, if I go back in history, every 10 to 15 or 20 twenty years—this goes back to the '60s where everyone thinks, "Oh my gosh, the computers are going to take over and they're going to replace everything." I think that it's always ended up being really helpful what they can do but never what people have imagined they are able to do, if that makes sense. But when it comes to the big question of," can computers replace artistic expression?" So, could a computer make music that could be indistinguishable from what I might write? I would not be surprised if that happens. You can already see Google has experiments where they have computers making paintings and there are some things that are very compelling.
But just because it looks the same, doesn't mean it has the same value as art. For me, art is a form of communication, so even if one piece of art looks or sounds similar to another, it doesn't hold the same intrinsic human story. As we think about the role that iZotope plays, our guiding principle is to expose the creative choices to our users and remove commodity, technical work. There are so many technical details that can get in the way of making art, we want to keep the artist in their creative headspace and not force them to switch to technical problem solving headspace. How many times have you been trying to capture a musical riff when instead you're re-installing drivers or tracking a sync issue that's causing clicks?
The important part that shouldn't get lost about machine learning is that it requires a huge amount of human interaction and direction. For our products, there's as much thought that goes into preparing and post-processing the audio with traditional signal processing that goes into the approach and architecture of machine learning components. It still takes an expert who understands audio production to setup a machine to get results which are meaningful. We're not at a place where a computer just improves on its own.
That's an excellent answer. Thanks Mark.
Finally a gratuitous pitch for Opcode.
Opcode Vinyl circa 1996
KVR Audio, Inc.