In an extension of iZotope’s video series, Are You Listening?, watch as legendary record producer, mix engineer, and author Sylvia Massy (Tool, System of A Down, Johnny Cash) and iZotope’s Director of Education and professional mastering engineer Jonathan Wyner (David Bowie, Aerosmith) discuss where mixing ends and mastering begins. They talk pain points in the mixing and mastering process and work through innovative ways to solve them in the box with iZotope’s powerful audio tools.
Jonathan Wyner: I'm Jonathan Wyner, Education Director for iZotope, and I am here in beautiful downtown Ashland, Oregon with Sylvia Massy in her compound called Studio-De-Lux. It's an amazing facility that's just full of interesting objects and tools where Sylvia does her mixing work. And we're going to spend some time together talking about mixing, mastering, and the relationship between the two.
I have some questions for Sylvia about mixing, and she has some questions for me about mastering. So let's dig in.
You started out as a musician, is that correct?
Sylvia Massy: Yes.
JW: Which probably allows you to have empathy for the musicians that you're working with.
SM: Exactly, in fact, I learned how to play drums just so I could understand the placement of the drums in a kit and where I could put microphones to not get in the way because I don't want to impede the way that the drummer plays.
JW: How did you learn to mix?
SM: It was interesting to start in the analog era because when you pushed up the faders for your mix you were really a part of a living thing. You could grab faders, several at a time, and whether you're using automation or not, I would do a lot of work live. That's my specialty. I'll put events into a mix that come out of my background as an analog mixer.
JW: Would it be fair to say that it's somewhat performance-based?
SM: Yes. Being that I'm not the primary musician in the project, this is how I can play.
JW: I don't know about you, but I've had this experience over and over again over decades where artists will start to do their own work and they'll get three months into a project and they're like, "Oh my stuff is shit. How come it's not getting better? It doesn't sound like my favorite record," and they don't get that it's a long-form process.
SM: Learning how to mix was an observation of popular music. When I first started, I would take my mixes and compare it to something that I was listening to on the radio. And I would do an A/B test and I would really learn to find the details of what made the commercial project finished, and then I would apply that to my mix. And it usually would take the shape of adjusting the overall compression and EQ, and so I was, in essence, doing some pre-mastering right from the beginning, because that's how I could quickly shape my mix to be more commercial and competitive.
JW: Is there anything that you know you're gonna do from the get-go in a mix, and at what point do you start doing it?
SM: When I was working at Larrabee they had GML 8200 EQs in the racks, and that was my go-to stereo bus EQ when I started out in analog mixing. And it was easy because I would just shape the mid range and I would scoop the mid range, and it seemed like a lot of commercial masterings would work that way. There was a little air on the top, a little scoop out of the middle, and maybe a little bump on the lowest frequencies, and it's very easy to just go bing bang bing and then do my A/B test, and I would get my mix quickly into a competitive form.
JW: So this observation that you used to make about the mid range being clear, or the top end having a little bit more energy and the low end having a little bit less energy and you take a look at an EQ that's sitting on your 2-mix for this track and there it is.
SM: Yeah, there it is.
JW: So is that something you start with at the beginning? Or you get X amount of the way into the mix and then do this?
SM: When I was working in analog mixing, I would go straight to that and GML EQ, but that wasn't until towards the end of the mix. But now, with template mixing, I will add this curve on to my stereo bus right away. I'll bypass it and just check and see how it's working with no stereo bus processing.
JW: So far, you've been talking about EQ. You have not been talking at all about compression. Is that also a standard part of your treatment?
SM: Absolutely. When I did analog mixing, I would have a selection of really great valve tube stereo bus compressors and I had a Fairchild 670 and I would use it on every mix, and it really finished the sound of the mix.
JW: The time constants in the vintage limiter are derived directly from the Fairchild 670, which is kind of interesting that you would pull it up, and especially in the tube mode.
Your comment about not being able to introduce as much level into a plug-in as you can in the analog world has to do with the fact that it's really hard to model the non-linear, chaotic behavior of an analog circuit in DSP. We can model the behavior when it's working within a "normal range," but if you want to introduce some unusual, aberrant behavior, that's harder to model and harder to predict.
SM: That's what I love, the aberrant behavior. The unpredictability.
JW: So this, for instance, represents what you would put on the 2-mix, as far as a treatment, when you're working in the box?
SM: So the way that I have this mix system set up is that my Pro Tools I/Os come out through these D/A converters, they're summed analog through an analog summing unit by Dangerous Music, and then they are put back into digital form, and I print it back into the session, so I can keep track of it. That's where they live and that's where I'll bounce out the files later. The final step before it's printed is where I put a stereo bus plug-in. So I'll have Ozone and I'll use the EQ and compression or limiting there.
JW: That hybrid approach that you're describing, that round trip through the analog world, does that give you a chance to insert other devices?
SM: This particular system has the option of me adding things into the stereo bus and I have a Western Electric 111 C, which is a pair of transformers that I'll put on the output.
JW: Can you describe where you're running into trouble with levels?
SM: Whether it's a session that I've recorded or a session that someone has sent me to mix, it seems like I start with levels too hot. I'll show you on this session: these are the levels that come to me on the Edit window, and then you can see that they're already brought down a little bit. This is just too hot already. When I play this track, if I was to bring this up to zero, you can hear the problem. So I have to bring my levels back before I even begin. And I'm a little bit afraid about harming the audio by reducing the level. Can you tell me about that? Am I harming the audio?
JW: Quite the opposite. I mean, certainly when you're hearing a distortion, that's an obvious sign that you need to change something. Right? So you're actually harming the audio by not changing your levels, right? So that's thing number one. I mean, it's funny, we always want to get the technical aspects of what we're doing right, but let's just start with the simple proposition that if something sounds wrong then we have to deal with that, to hell with the bits, right?
It makes sense, certainly to me, if you're recording all of your tracks at a level where the peak is somewhere, say, between -3 and -10, right, for every individual track. Once you've summed a bunch of channels together, like what you see here, inevitably you're going to end up with a level that's too hot, and you're going to have to turn it down.
The good news is, with Pro Tools, Studio One, or any sort of modern DAW, we're operating in a floating-point digital universe.
So long as you're not making changes that are in the order of 40, 50, or 60 dB, then I start to hear then change that happens because I'm making extreme changes in gain with floating point you've got so much dynamic range.
SM: So I could actually take these and start at even a lower level? Like even a -30 and be okay?
JW: That's correct.
SM: And now, by doing that, I have so much more room to work with on my stereo bus. That's good news because that is a constant issue. I'll have a client, I'll get the mix just right, and just on the edge and then the client will say, "Hey, can you bring up those guitars?" and the drummer will say, "Hey, can you bring up the drums?" and the vocalist will say, "Hey, I can't hear my vocals anymore." You know, everybody wants to be louder. Pretty soon we've lost the headroom when we're distorting again and then I have to rebalance and bring it all back down.
So it's this constant fight. And I want to play with that more, too, I want to play with the stereo bus and try to push the stereo bus harder instead of having these loud tracks.
JW: There's an art to the compromise between level of your individual tracks or your stems—some people mix to buses—and then that gets combined to a 2-mix—
SM: That's the way I'm doing it here, too.
JW: So the level there and the level on the 2-mix: If you leave too much for the mastering stage to get the level that ultimately the artist, the client, or you, yourself are going to desire, you won't be able to achieve it without having to be more assertive with a 2-mix processing, or the mastering processing, than you would like.
The mix engineer, in my way of thinking, has to pay attention, to some extent, to the difference between the peak levels in the average levels. That's sort of a big deal.
SM: The other thing is, I always thought that there would be a global adjustment, maybe in the I/O, that we could just bring everything down and start from a lower level, but there really isn't. So what I was doing was adjusting the level on the stereo bus through using the Ozone inputs and outputs and then I would reduce it on the input and output. But it wasn't solving the pinched sound of having everything just on the edge, so the solution for me is to bring it all the way down to -30 to start. Whereas, people give me .WAV files and they start at zero.
Before I start mixing everything's brought down to -30, and now I'm having success. I just was really curious if I was doing something wrong, so thanks for that.
JW: Sure. Like you are sort of fond of saying, there are no right answers, there are no rules here, so these are all guidelines. But in the case of the mix that you've got going here, if I hit play and we just observe the levels on Insight and take a look at the average level versus the peak level, I think it's somewhat instructive. So I'm going to just play a little bit of audio and then we'll come back and talk about it.
A couple of things are going on here. The average level in the loudest section of the tune at the end was sitting around -6. If you think about the old-school sort of VU meters that we grew up on, usually we would target something between 0 and +3, but that zero on the VU meter was usually about 20 dB below clipping. The way that we used to make mixes that we were happy with, the level is much lower, so here, this is quite hot.
So I would venture to say if we leave this mix at this level, you're going to get a plenty-hot record. There will be no problem, in terms of the level, that you're going to get in the final result.
SM: But the mastering engineer needs some headroom to work with.
JW: That's exactly right. There are a couple of ways of thinking about this idea of headroom. One way of thinking about it is, "Well, the mix sounds great, let's just turn it down a little bit." And if the mix truly sounds great, if you're totally happy with it, and you have metering that shows you what we're seeing here, then, whether it's you or the mastering engineer that just pulls the output down, that final version, to get some room to work, that's an option. I'm not saying it's the only option, nor is it always the right option.
If you're saying to me. "You know what, I think it sounds better if I give myself more room to work so it doesn't sound pinched," I, as a mastering engineer, would be very happy to have some more headroom above where the average level is sitting than what we have here.
SM: Well this particular project, the mixes were all the way up to the edge. Then, when it came to mastering, the mastering engineer had difficulty to keep it clear because the guitars were obscuring the vocals. Things got lost right away. I went back three times bringing down the levels just to get it to open up enough that when I got the mastering back it made sense and it wasn't worse than the final mix. Understanding how far down I can bring the initial tracks is going to help me a lot with my future mixing.
Now this meter is really revealing, and I wanted to ask you, where do you have that? Is that post-Ozone on the stereo bus?
JW: Yes. In order for that meter to truly be useful, it needs to be the very last thing in the chain, because it'll show you the accumulated level through everything that you're doing, whether it's on the track level or the stem level or the 2-mix out. It's a complex system, this sort of gain staging through a mix, right, you've got your stems your busses, right. Do you have anything on your buses?
JW: So, you know, one of the things you have to be aware of is that if you drop the level of the tracks it's gonna change the processing so you have to do all of that again. And that pinch phenomenon I think that you're alluding to comes from pushing level into your 2-bus processing into the limiter.
SM: Oh really, is that right?
JW: Yep. You probably would get some satisfaction if you drop the input into Ozone, but still, anything that's happening upstream from that, in terms of gain staging, is still going to be problematic, so yeah feel free to take the tracks down, give yourself plenty of room to work, and then you can add level later. It's really hard to take level away and undo some of the artifact and some of the things that you're hearing.
There's one other thing I wanted to point out: so you've got a limiter here that's your final stage. This is sitting on your 2-mix output, and then the ceiling is sitting at zero. I always encourage people at every stage to leave a little bit of margin coming out of their final stage. It gives you a little bit of room here so you won't get levels that peak above 0 dBFS. You won't hear distortion on the output if you if you let this sit at zero. There's every possibility that on the analog side what we're hearing actually will exceed zero and it'll sound a little bit clipped. So whatever your final stage is, go ahead and cheat that down just a little bit, give yourself a little bit of room. It'll make mastering engineers happier too.
Anything more than a dB or two is not gonna help, again, the pinched thing that happens because of the gain staging into the limiter. But it's always a good habit or a good practice to leave a little bit of margin at the very top for when you print your files.
SM: Is that gonna create the pinching sound?
JW: Turning it down? No, absolutely not. It's just a function of level. If anything, it should make your peaks sound cleaner and better and clearer.
You know, a lot of people are having to kind of master their own stuff, so people start thinking about mixing and mastering at the same time. And so I just wanted to get your take and your approach to that issue.
SM: When I start on a project from day one of recording I'm always thinking of what it is going to be when it's mixed, and, ultimately, mastered. So I make placements in monitoring, while we're monitoring, while we're recording, I'm combining things, I'm placing them in the panning, I'm doing an EQ on the way to the recorder so that I don't have to make these decisions later. And ultimately, those things are meant to make mixing easier. So it's already thought out when I go to mix.
The trick that I've found is to get an approval from a client. That mix has to sound competitive with what's on the radio, which has already been mastered. So I'm always going to do some pre-mastering just to get the client happy and excited and so we can really actually understand what it'll sound like.
So I want to start right away with understanding where the mix is going to be and then, ultimately, the mastering. Once I get an approval, then I'll strip off the stereo bus compression. I'll use it for the approval, and then I'll take it right off before I send it to master.
JW: Talk to me about your relationship with the mastering.
SM: I just don't want them to screw it up. When I send them, without processing, without stereo bus processing, I also send them the reference, the final approved mix with the processing, so that they have a reference. "This is what the client approved. Please, if you're going to do anything, make it sound a little better than that. If you're going to do anything, just don't make it sound worse."
What do you think the job of a mastering engineer is? The title is "engineer." Does that mean that they're being creative with what they do?
JW: That's a hard question to answer. You use the word "creative," I think you have to tread gently with that. Sometimes creativity and mastering, to me, may be the gentle addition of color that might come, in some cases, fairly transparent. Transformers, and very rarely tape sort of treatments or something like that, can benefit a mix, but only a little bit. And if you go too far, you change a lot about the way a mix is put together,. You can change the balance and you can change the tone. I think the responsibility is to sort of notice what you notice about what's going on in the mix, if you have any questions communicate with the mix engineer and say, "Did you really mean for the pennywhistle to be 30 dB louder than the vocal?" and if they say, "Yes," you have to figure out a way to make that work.
It's not your job to know what's right. It's your job to make the observations, and then to help. I think the key words are "respect" and "a little goes a long way."
SM: I find it amazing to see what a mastering engineer can do to broaden the width of a mix, but that often is at the expense of certain levels of things that you've placed in the mix. When I get a mix back from mastering and I'm listening to it and now it's wider, it's airier, it's cleaner, but my lead vocal is buried behind the freshly exposed guitar parts, so what do you do?
Well I have to go back in and I'll make a vocal up version, or anything that's obscured by this new finishing EQ. If I like it, I'll make adjustments to my final mix and send a revision over to the mastering engineer so he can remaster using the same settings, just will just be able to hear those parts better.
JW: That's very cool. So you hear something that's revealed in mastering that you do like, and you sort of want to work towards that goal.
SM: Right. I used to work with Prince, and he would record in a way that made me uncomfortable because he was pushing the levels on the tape so hard that if the monitors weren't up in the room, you could hear on the tape machines the rattle of the needles because they were all pegging, and there was a sound to that. If you listen to those Revolution records and anything that he recorded, and he did all the recording mainly himself, they had a sound, it was smashed and really in your face, so I love that. But add a hundred tracks together and you've got this crushing mess.
Now, when I start a mix I take every track, before I even begin, and bring all the levels down, whether it's something that I recorded or if it's something that was sent to me to mix.
JW: That looks to me like what I would hope for from a mix engineer. There's a little bit of a margin at the top, there's a little bit of room to push level if it needs to be pushed, but there's also enough average level that it's not going to be up to mastering to make the record sound like it needs to sound.
There's something else that we noticed that was really interesting about this record. If we pull up Tonal Balance Control, which is this tool that's designed to give you a sense of what is—I hate to use the word "typical" or "common," because that just sounds really uninteresting, like "I want to make a common record or a typical record." Every record, of course, is supposed to sound great and compelling and interesting and beautiful. But if you look at the display, you see a range what we like to call "tonal balance," right, low ends to mid-range to top end, the average distribution of energy across thousands and thousands of records.
SM: So this is machine learning that's evaluated thousands of records?
JW: Correct. But it doesn't mean that this is the final answer, nor does it mean that you have to make a straight line image of, you know, in this display in order to have a good record, so when we hit play and let this settle down for a moment, you can see, and this is this is a fairly aggressive record, so you might expect to see the bottom and the upper mid range being a little bit more emphasized, and, in fact, it's still kind of within bounds. It's not like anything is so whacked that you'd have to go back and revisit your decisions, or else you'd blow up a car speaker or something like that right?
SM: Well that definitely illustrates the scoop that I pull out of the stereo bus on this genre of music, because it's a it's a harder, heavier genre, and it's kind of a scoop.
JW: Right. Actually that's right, so if we take a look at the Ozone EQ, you can actually see that right there this influencing that shape.
SM: How about that? That is an amazing tool.
JW: You and I have had the luxury of honing our craft in very specific parts of the music workflow. Not everybody has the luxury of doing that. Some people love doing it all themselves. If you're talking to somebody who wants, for instance, to master their own mixes, do you have any advice that you would give them?
SM: I wouldn't push anyone away from doing that, actually, because you can get so much out of just taking your mix and A/Bing it with someone else's work, and really dissect what is the difference, what makes it sound good, what is the ultimate shape of that that successful mastering? And you can do that yourself just with a few tools. iZotope has a great suite for mastering, so I would suggest everyone try it at least.