A much-in-demand mastering engineer takes you to school on the mastering process
Interview by Paul Vnuk Jr.
Stephen Marsh has a discography that goes on for pages and reads like a Who’s Who of the entertainment world. In addition to mastering everyone from Jars Of Clay to Keb’ Mo’, Ben Harper to The Donnas, and Jeff Beck to Los Lobos, his mastering expertise has graced dozens of film and TV soundtracks like Veronica Mars, X-Men: Days Of Future Past, Elf, and Life Of Pi. We had a great chat with him and learned his history, his methods, and his suggestions for readers who might want to try mastering themselves.
- On Mastering
- Q&A With A Master: Howie Weinberg (v2)
- Mastered For iTunes: High-Resolution Audio?
- Stephen Marsh on Magic in Mastering
- Q&A With A Master: Howie Weinberg
- Q&A With A Master: David Glasser
- Q&A With A Master: Brad Blackwood
- Blue-Collar Mastering: Interview with Chris Sonia
- DIY Mastering
- Preparing for Mastering
Can you summarize your background? Where did you get started and how did you end up with your own place, with your name on the door?
Stephen Marsh: I grew up in New York and I came out to LA when I was 19, knowing nobody, and I answered an ad on a bulletin board for a studio looking for a runner. That was Sony Music Studios. Within about three or four months of being a runner, they let go of the night guy that worked out of the mastering room, and then said in the proverbial, “Hey, kid, do you want a shot at the big time?” It was just running tape clones in the middle of the night, but it was cool in that I got to scrutinize and listen to fully prepped digital album masters of every top engineer that was doing work for Sony Music Labels at the time.
For the most part, digital cloning consists of data checking, so you can actually run it through a desk and run some EQ and experiment with things as it runs. I could listen to the night’s playback of the new Pearl Jam record and learn how to hear and how to understand just exactly what mastering was and what it wasn’t. I found that I sort of had a proclivity for it, so I apprenticed under a guy named David Mitson and worked for him for about seven years.
At the end of 2001 Sony closed the West Coast operation; two other guys and I bought out the whole facility and the staff, and we set up a new facility called Threshold. Then I split off the mastering operation from that. That was about eight years ago, I guess, and that was where Marsh Mastering started.
So did you do any traditional engineering along the way?
I’ve kind of always been a mastering guy. I played in bands growing up, and I was always the techie guy in the band that ran the PA, so that was my experience with engineering. I was very fortunate to be able to turn that into a career.
You are the fourth mastering engineer I have interviewed over the years, and only one of them—Trevor Sadler (December 2007)—actually started as a tracking and mixing engineer. I interviewed Howie Weinberg two years ago; he started as a mastering guy, and it’s the same with John Davis, who just remastered the Led Zeppelin catalog.
I have all three of those first Zeppelin remasters sitting about three feet away from me right now. I was really smoked by what they did and I’ve listened to these records since I was a kid. I know everybody says it, but I heard new stuff on that that I’ve never heard before, and I dig that!
What’s your typical workflow? Do you start with a certain EQ, or a certain signal flow?
Like a sonic recipe? It’s kinda weird ’cos I don’t. I want to say jokingly that the way that I play a mastering console is the same way I used to play the drums—I sit down behind it and I react musically to what I hear. It’s all second nature. One of the beauties of working in a room where you know the gear really well is you don’t think about “how am I going to do what I am doing?”, instead you just think “what do I want to do?” and mastering becomes possibilities, not challenges.
So what is Step One when you start a new project?
We work everything at high res here. So even if it comes in 16/44.1, the first thing I do is bump it up with an algorithmic process to a higher resolution, typically 24/88.2. This way any digital processing that I do gets twice the mathematical resolution and it’s captured back at a higher res too after going through the console. Typically we work out of Pro Tools and sometimes we use Prism converters, sometimes we use Burls; we used to use Apogees as well.
All of my EQ happens before my compression. I don’t know if that’s “right” or “wrong”, frankly, and I don’t really have an opinion—it’s just the way I’ve always worked.
That seems to be the general consensus I have found with most mastering engineers I know, EQ > compressor > limiter.
That’s ostensibly what I work with these days. I had a much larger, more elaborate console in my last place. I’ve spent the first ten years of my career adding gear, and I’ve sort of spent the last ten years of my career peeling it away to just the things that I need. Sometimes you just want to say, “Well, it needs 8K, and here’s the 8K that I got.” The limits sometimes empower you.
We recently pulled all the low res digital stuff out of the desk. If it doesn’t go up to 24/192, I just don’t want it in the room. I don’t want bottlenecks, I want to do all the work as dynamically as I can. I’d rather have everything go up to DSD, but I’m living in a fantasy world in that regard, I suppose. [laughs]
I do all my EQing before compression, an EAR tube EQ first, then a discrete GML EQ. The GML is also wrapped through my Mid/Side network and can go either Left/Right or M/S. Steve Firlotte from Inward Connections built me a modified EMI spreader control, that’s in the chain. I have two different elliptical filters for summing to mono, and a Maselec limiter. For an SSL-style compressor, the repeatability, the infallibility, and just the clean gain you can get out of the thing is astounding. I love it.
How is this all tied together? What is your mastering console?
It’s all housed in an Argosy desk and the center section is custom built by Steve Firlotte from Inward, I would say it’s his version of Avocet or Dangerous, though this design actually predates both pieces. He’s also done all my mods. All the gear is [wired] point-to-point with no patch bay. There’s one main XLR patch point to get into the front of the desk and there’s one to get out of the desk. Anything that needs to be bypassed has switchable relay bypasses, and there are VU meters on the front end, VUs on the back end, and then VUs on the main program. So I can monitor pretty much anything that’s going on in the studio, from any position.
Most of the desk runs on DC; there are a couple of pieces where the power supplies are still in the units, the Burl converter mainly. Everything else has been remoted, just to keep it clean, and most of the power supplies sit in another area. It keeps the room really cool; we run on almost no air conditioning here.
We do have a Studer and an Ampex, and if the analog decks are cranking doing transfers it gets a little warm. But it gets a little warmer any time we flip on technology from 1976!
The only other piece of analog gear I should mention, which has just been added to the desk, is an [elysia] xfilter.
I reviewed that and really liked it.
I didn’t have a device really for filtering low frequency rumble. If it’s PCM work I can do it in the box, but if it’s DSD I do not have the ability to process that whatsoever once I’ve captured it. I’m still learning what it can do, what its strengths and its weaknesses are, but it’s fun.
So what do you capture back to?
If it’s PCM I capture back to Pro Tools, if it’s DSD I capture to the Merging Technologies Pyramix.
Let’s talk plug-ins…
I’m pretty old-school when it comes to plug-ins. I still use a Waves L2 on occasion.
Gasp! This interview is over! [both laugh]
I get most of the gain that I need from my desk through gain staging and trans-formers, so I really use it just to keep from getting overs. I’m using like six- or seven-tenths of a dB, which is typical of the way I used the hardware.
I had a hardware L2 in my desk for years, back when it first came out like in 1999 or 2000, and as the level wars really kicked in, we started getting 2 and 3, sometimes up to 4 dB out of it. That’s just too much; you start to suck all of the dimension out of the song. Not that it’s a fault of the plug-in, it’s just that’s what it has to do to give you level for free, basically. Now we’re back to using it very minimally, and it works in that capacity. I also have a couple of other plug-ins I’ve used from Massey and iZotope, and I have the elysia alpha compressor.
I use that one, it’s really gentle for a plug-in compressor.
I am still experimenting with it. I have trouble knowing how far to turn the knobs on a plug-in. I can turn the knob on a piece of analog gear and hear it directly, but a lot of times the plug-in will have functions or buttons that I don’t inherently know what they do. It’s a lift button, or it’s a shunt button, or it’s a floor button…
On hardware, when I adjust a lowcut filter and roll off bottom end, I understand that it’s pushing up one peak by making dips around it. I don’t have that shorthand with plug-ins in many cases. So I don’t know how far to turn things, and I have to experiment a lot.
What’s your monitoring chain? Do you just work with one set of monitors or do you have multiple sets?
I have three. I have PMC mains, IB1s, that I’ve had for a long time, with a pair of Bryston 7BST amps. I have a set of PreSonus Sceptres that we use as midfields, and I also have a set of Avantone Mixcubes. Basically, I’m spending 80% of my time on the PMCs.
Mixcubes in a mastering room? Usually I see those in mixing rooms.
Once people started listening to music through their mobile phones, laptop speakers, and through headphones, it became imperative and incumbent upon any mastering engineer prepping things for commercial distribution to know what it sounded like on range-limited single-driver systems.
What about musical genre? Do you have a specialty? Are there any styles where you say, “Yeah, that’s not for me”?
Genre-wise, I’ll work on just about anything. Growing up I went through every phase you could name; I listened to a lot of rock, hard rock, metal, rap and hip-hop… I never got way into dance music, but I listened to Top 40; I listened to what was on the radio and I always have. I would study and listen to jazz music.
I’ve done everything from Peter Paul And Mary to Skrillex, and I’m kind of comfortable working on just about any genre. If I’m unfamiliar with a genre or subgenre with intrinsic sounds and sonic signatures, I can be taught. I always consider my job as like a sonic translator.
Predominantly my work these days is with scores and soundtracks, so I do a lot of orchestral and compilation work, which incorporates every genre known to man, depending on the film. And I do a lot of remastering where we are doing predominantly SACDs. We’ve also mastered high-res digital for vinyl cutting and that’s becoming increasingly popular.
Digitally sourced vinyl seems to be a pretty hotly-debated topic on the forums right now.
So much really good vinyl—whether people want to accept it or not—is being cut off high-res digital these days.
When I spoke with John Davis about mastering the Led Zeppelin vinyl, I brought up the question of cutting right from the original analog tapes, and he mentioned that you couldn’t cut vinyl from them. There are dropouts and bad edit points….
Bumps, wows, jumps. Yeah, absolutely right.
One of your jobs is remastering well-known and loved classic albums for Audio Fidelity (audiofidelity.net), where you work with re-issue producer/mastering engineer Steve Hoffman. What’s that like?
In that situation, he really focuses on the production and the creativity. That allows me to be much more nuts and bolts, and just operate as an engineer. He will call out a frequency and say, “Hey, give me 2 dB at 50 Hz” or something, or “I think it needs this and this,” and I execute those basically the best way that I see fit across the desk. So instead of coming up with the recipe and then tasting it, I’m simply throwing the ingredients I’ve been given together in the right amounts and then I’m tasting it! It’s a different experience for me and I thoroughly enjoy it.
Do you approach remasters differently from clients with current projects?
We don’t use any compression, and there’s nothing with an “-imager” or an “-izer” or an “-esser” on it in the signal chain. It’s just EQ, gain, and the sound of the desk. So in that sense our palette of what we can do is very limited.
Occasionally, you have to do other histrionics. If you get a tape where the master’s been edited from multiple takes and it’s off analog tape, you’ll probably have a slightly different azimuth character on every single piece of that tape. Then you have to do it in chunks, because you have to align the azimuth for each of one of those little sections and then put it back all together. That can be fun. Not!
Have you ever had moments similar to what John Davis describes with the low frequencies of the bass and kick drum found with the Zeppelin tapes, where you put a master tape on and hear something you have never heard before? If so, do you leave that in, or do you and Steve prefer to get it sounding like the vintage vinyl?
Oh yeah. We hear all kinds of different things! Steve really takes each master tape as it comes, and he treats them as they are, but references are always available.
Ahead of time I’ll load in previous CD releases for reference. If there’s been a previous remaster we’ll try to locate it. The labels we work with are very good; if there have been been four releases of it already, they’ll be very good about trying to get us four releases. They’re also really good about getting us original first-pressing vinyl of almost everything from the vinyl era that we do.
So I listen to all those, and sometimes I’ll load in some samples and during a session I will point out if there’s something askew. Like, if every CD release is really bright but the master tapes are really dark, he’ll listen and say, “Is it bright because it was emphasized, is it bright because the original A/D converters were bright? Is it because it was 1983 and everybody was high on coke… is there something to explain it?” [laughs]
The fun thing for me on the original master tapes is when you hear the ends of songs without the fades. It’s like, “I never knew there was another eight bars of solo after that!” It’s fun to hear those kind of things sometimes, and occasionally we’ll be allowed to leave that kind of stuff in, but in most cases I have to fade it to match the original release.
The other thing we do specific to vinyl-era reissues is that we want the listening experience to be similar, so we will actually put six or seven seconds of silence between what was Side A and Side B. And that’s to honor the artist’s original vision; the artists never intended Side B, Track 1, to be played two seconds after the end of Side A. There was meant to be a record flip. So, in a sense, you kind of want to treat them as halves. That’s me saying that, it’s not necessarily Steve. It’s sort of my rationalization for why we do it that way, and I think it rings true.
Let’s move onto some practical questions. What are some of the common mix mistakes that people are doing now that drive you nuts?
Like sonic atrocities? I find, if there’s any one constant problem, it’s not understanding the concept of level and balance. Mixing is inside-out listening; you’re taking all the individual pieces and creating the whole. Your two main tools, before anything else, are level and balance. Where does something sit across the mix? Is it here, is it there, and how loud is it in relation to everything else? Then maybe you add a little EQ to make something pop, but that’s all secondary.
Mastering is listening from the outside. Anything I do affects the whole mix. Mastering is the completion of the process of mixing. They are not two wholly separate things; to my way of thinking, they are two separate functions of the same thing. I think in most cases, if I’m getting problems with the mix, it’s because they haven’t understood that mastering is the final part of mixing and it doesn’t need to sound “like a finished record” before it comes in here. In fact it shouldn’t, because if you push it right up to the edge of the line, there’s nothing left.
So do you have a level that you recommend? Should a final mix leave at least 6 dB of headroom, 12 dB, or even 18 dB?
Stay away from the last bit of your digital converter and that’s usually 6 dB. Basically, if you’ve got 6 dB of clear headroom in your mixes, you’re in good shape. If I need to adjust the output levels, I can adjust the output levels. I don’t necessarily need any headroom; that’s kind of a fallacy, but I think it’s a fallacy that remains in place for a good reason. If you stay away from that top bit of your converter when you’re capturing it, generally you’re going to be a lot safer and have a much cleaner sound.
Do you prefer your sessions attended or unattended?
I’ll work happily either way. For unattended, I will generally spend a little more time on those projects because I don’t have anybody in the room to read. I will say that the time that I stop is when I get goosebumps! For attended, there’s something with creative human beings in the same room together… anybody that’s been in a studio that’s felt that magic before, understands it. It’s like a certain point where you don’t have to ask the person sitting on the couch behind you if they like what they hear—you can feel it.
To wrap this up, I wonder if you could address the budding mastering engineers out there with Stephen Marsh’s top three dos or don’ts.
Yes! Two dos and a don’t…
First, understand your monitoring, and I don’t just mean what your speakers sound like. Understand what they sound like in the environment in which you’re listening to them. Listen to music that you like the sound of, and listen a lot! Then listen to other stuff, too. Because music in a certain genre is going to have a certain type of character, and it’s not necessarily going to reveal flaws in a speaker system until you’ve listened to a lot of stuff in other genres.
Second, if you’re going to buy gear, start with what feeds your ears and then work your way back. When you do that, you find out that the last thing you wind up buying is plug-ins or processing hardware! After monitors, focus on converters next. Then focus on the processing. Every piece of equipment, every plug-in, is going to have its strengths and weaknesses. The magic of mastering is finding just the right pieces that do what you want to do, finding exactly what they do best, and only using them for that. Make sure everything in your room works flawlessly together, so you’re not thinking about what limitations you’ve got, you’re thinking about what possibilities you have.
Third, don’t do too much! Understand that a lot of the best-mastered albums, that are revered, have one band of EQ on each song. Listen to what a song needs and what it doesn’t need, and only do the bare minimum that it needs to really make it sing. Oh, and when all else fails, try stuff! Don’t be afraid to try stuff, that’s how we learn! Listen for the goosebumps… and then stop!
Learn more about Stephen at his web-site, marshmastering.com.