Michael Romanowski: Mastering, AI, and embracing imperfection

With five Grammy awards to his name and client list that includes Alicia Keys, Tom Petty, The Bee Gees, Al Green, Lady Gaga, Tears For Fears, and more, Michael Romanowski is one of the world’s leading mastering engineers. Likewise, his California studio Coast Mastering is one of the industry’s most sought-after facilities. Here, he joins Headliner for a chat about his route into the industry, the rise of AI, and the true art of mastering…

Tell us about your background in music?

I was in Mexico City with Lesley-Ann Jones recently at the Soundcheck Expo playing some files and sound specs. We had just finished working on some cartoon music from early Warner Bros stuff and it sparked a memory for me, as I realized that even as a kid I was aware of the impact music and sound had on the visuals I was watching.

I got into playing music when I was young, and I joined a band in college playing bass. One day we were playing at a local club, and I noticed that there's somebody behind a desk twisting knobs. That was the first time I realized there was somebody doing sound for the people onstage. I would ask questions and I kept bugging the guy and eventually he’d let me take over while he was having a break. Then it became, ‘I'm taking the night off, you run it’. And after a while it became, ‘I quit, you're the house sound man’!

Around the same time, the band was getting into recording. And I realized the recording process - the console, the listening, the miking – shared some similarities. So, I started recording the band, and then other people’s bands… it just came along in that way.

So how did you get into mastering?

I was at a point in my career where I took a little break to drive around the country for a month on my motorcycle. One day in San Francisco, I found I had a message on my machine from a mastering facility in the Bay Area. They said they were looking for another engineer so would I send a resume? I called up and said, ‘I'm on a motorcycle by myself so I don't have a resume, but I am in town. I could just swing by’. So, I did. But even at that point, I didn't know what mastering was. I just knew it was something that needed to be done. Back then it was much more mysterious.

The place was called Rocket Lab, and when I got there, I met Paul Stubblebine and Ken Lee, both fantastic engineers. Paul took me under his wing and started showing me what mastering really is. I found it fascinating. I hadn't gone seeking it, I just followed an open door. They offered me a job and now it's been 30 years.

Is there still a sense of mystery around mastering? And does that lead to a lack of appetite for mastering engineering roles compared to recording or mix engineers?

There are a whole lot of answers, and I am going to get on my soapbox for a minute about the state of the mastering industry. When I got into it, it was definitely considered an art form. You would be an apprentice, working with someone who knew what they were doing, and you’d work with them for years. That’s how I learned. That's how my heroes learned.

These days, there are some who want to get into it for the right reasons, but some think they just need a plugin or a program and they can become a mastering engineer. They think they can put something on the back of a mix bus and just call it mastered. That isn’t the art form of translatability. It's not understanding the manufacturer. It's not understanding distribution. It's not even understanding what the artist wants.

And AI mastering is the dumbest thing in the world. The whole point of AI is to take a subset or take a set, a learned model, and find an average. And the larger the model, the better the average. And if you are trying to make music average, go AI. Let AI be your mastering tool because you're going to race to mediocrity.Music lives on the fringes. Art lives on the fringes, pushing the boundaries, trying new things.

Tell us about your approach to mastering?

The technical stuff is things like providing the right master for the right distribution. But there is also the artistic side of it; the 10,000-foot view of how is it, not what is it? And the people who are trying to put plugins on the master bus or whatever, are still thinking about what it is, not how is it. We need to be up here at the 10,000-foot view. Like, this is how this album feels. This is a body of work, not a collection of songs. This is a vibe and a feel and something that needs to get across.

Has the demand for things like AI/plugin mastering solutions been driven by tech or the lack of money being earned by artists today? Or both?

The technology has allowed more people to make records. People curious about the technology might want to know the process and do as much of it as they can themselves. It's not unlike cooking. I can go to a shop and buy a really nice set of pans, and I can watch videos and learn how to cook. That doesn't mean I'm going to open a restaurant. But I can get into the details to figure out how to cook and make a good meal.

And there is the issue of artists not making enough money. Less music is being sold; streamers are paying less. And then they can't afford to hire people like me, and being a mastering engineer, you’re at the back of the food chain.

With things like AI, what we haven't really gotten into is the mindset of why are we doing this? If we're using things like AI for creativity and pushing boundaries, great. Music is a way of communicating emotion from one person to another. The imperfections in that expression are the things that I gravitate toward. Listening to The Faces, something like that from the ‘70s, there are mistakes all over it. But man, it's fun. It rocks. They're having a great time. It conveys something. It feels great. It doesn't have to be perfect. If you autotune Chet Baker, he wouldn't have been Chet Baker. I somewhat long for the days of imperfection.

Tell us about your relationship with Merging Technologies - they have long been a part of your studio setup.

With early workstations there were definitive sounds. You could go, ‘that's Nuendo, that's Pro Tools, that's Pyramix, that's Sonic Solutions’. And Pyramix sounded great from early on. And it’s the same with the hardware. The converters were always really great sounding. They were the first as a system to be able to do DSD, you could edit natively without having to convert.

What we want in the mastering world is things that are as transparent as possible. I don't need my equipment to color the sound. I want to color it myself. I don't want the math involved in the bus structure of a workstation to affect the sound. It's one of the things I found about Merging stuff right away. It’s super clean. What you put in is what you got out. When making these decisions about how some of these masters are going to translate, I don't want it sugar-coated. I need it honest. And that's what Merging always was and always is.

I'm currently using two Anubis. I have two rooms tied together and I'm using two of the Hapis with 16 channels of mic pre and D to A on each one. And then two Anubis’ are running the system. I also have one Neumann MT48. I was really happy to see that Neumann bought Merging. Merging has a great product line, and Neumann has a great history of working with clients, and putting the two together was really great and advantageous.

So, I have the Hapi and the Anubis in both rooms, and my Pyramix system is hooked up as a source to the Anubis. It's a good brain. It allows me to hear, especially in the immersive world, different headphone mixes, where I can listen to the binaural versus the stereo very easily. The converters sound fantastic.

And as a portable unit the Anubis is great for doing field recordings, stuff like that. You can toss it in a bag with a laptop and make some excellent recordings. So yeah, kudos to them and their hardware!