RICK ROWE: CONVERSATION WITH A MASTERING ENGINEER

You’ve been working on a CD for the Rock & Roll Hall of Fame. Can you tell us a little about that?

It’s a compilation project. It’s a compilation of other people’s work. So you’re limited in what you can do but you have to do something because there’s no way to take 21 pieces of music from 21 different engineers and 21 different studios, put them on the same CD and think they’re actually going to sound good together. So changes have to be made but they are very subtle and it is really walking a tightrope of not trying to undo what another colleague of yours may have spent hours or days doing. Also, one of the most difficult things to master is a compilation of famous stuff that people know so you really can’t change it and yet you have to. It’s like a beauty contest in a way. We’re looking for who’s closest to perfection and we’re not going to mess with those. Now everybody else has to come up to that standard.

There are a number of programs for computers that allow someone to master on their own. Why would you want to hire a “mastering engineer” to do this for you?

A lot of times there can be errors that you don’t hear because you’re too close to it. It’s like the forest and the trees thing. After a while you’re only seeing individual trees when you’re really trying to create a forest. It’s hard to keep the concept of the forest when you’re working at it every day.

Mastering programs are a bit like pre-cut clothing off the rack. They’re generic, they’re going to solve generic inaccuracies. Proper mastering is a lot like a hand-tailored outfit, specifically created for you, or in this case, for your mix. And mastering properly, by hand, renders a universally worldwide playable product. And you have to think worldwide because you’re thinking about the Internet. You’re not thinking local. You’re going to be selling records to people in Lithuania, to people in the Philippines and Australia. So you want your CD to be playable on any kind of system, from the cheapest to the most expensive. That’s something a mastering engineer is aware of and can compensate for.

If you think of mastering as a meat and potatoes dinner, you can get it at a high end steakhouse, or, get a burger Mac Donald’s with a sack of fries. They’re both meat and potatoes but there’s a world of difference. And that’s the difference between using a mastering engineer and using a mastering plug in. And if that meal is going to represent your talent, you probably want to go to the high-end steakhouse. Also, a good mastering engineer will make himself available to any QC person at any duplication facility to answer questions abut your CD which can be an invaluable interface.

How did you get started?

I got started in 1970 by taking a summer job at Electric Lady Studios and I was having a ball. The Chief Engineer was Eddie Kramer, one of the most brilliant engineers of our time. I was an Assistant Engineer. At the end of the summer I went up to Eddie’s office to let him know that I had had an enormously wonderful time and to thank him for all the knowledge he had imparted but I was going back to class (I was a student at Mannes School of Music). But before I could tell him that he said to me, “You’re showing great promise and I’ve got Malcolm Cecil and Bob Margouleff coming in and I want to put someone with them and I think you’ll be fine for that. You’ll be starting work on a Stevie Wonder album in about two weeks.” So then he said, “What did you want to see me about?” And I said, “Great cravat.” He had a great collection of cravats that people wore during the day. So I said to myself, “Back to school or Stevie Wonder, back to school or Stevie Wonder?” It was a no-brainer.

Both Bob Margouleff and Malcolm Cecil were wonderful and I learned so much from them. And that album turned out to be Music of my Mind which was a great departure for Stevie Wonder at that time.

How did Stevie Wonder record?

It was very interesting. He had a basic set of lyrics that he would have worked out and those were written down. And Stevie would be out at the Clavinet because we would lay down the basic tracks with only the clavinet and Stevie. The way he was churning out songs he would be stopping us when we were setting up for an overdub to work on another idea for a song. So we would be throwing on fresh tape again because we were going to cut another basic. What would happen is that Stevie would have a headphone on one ear only and Malcolm would be in the studio with a microphone sounding like Whispering Joe Wilson and he would be feeding Stevie the lyrics one line in advance. This happened so often that I still remember to this day that by the time we finished with all of the overdubs, even before mixing, we had 43 songs in the can. And many of those went on to be used on Talking Book Part One and Two. It was an incredible thing. You couldn’t stay at Electric Lady. There were the staff engineers but when you came of age they would wish you well and send you on your way with the highest recommendation.

So I worked at a number of studios and worked my way up through the ranks for a while. One of the places I went to was Sound Ideas where I got to meet Leonard Cohen. I did an album, New Skin for the Old Ceremony”, with him and was asked to go on his European tour and mix front of the house. And I jumped at the chance to do that. It was basically an acoustic tour, a mixer’s dream. It was a lot of fun and quite successful. All in all I think I did a couple of US tours and 3 European tours. It was a great learning experience doing things live. There was no room for mistakes.

When I came back from one tour I discovered that I had a Disco hit, A Fifth of Beethoven, which I had done at Sound Ideas. I had forgotten I had done it but it was Number 1 in Billboard for three weeks and I got a Gold Record for that. It was done with real guys with real instruments. Thirty some odd string players would come in and twenty some odd cigars would be lit. There were some great players.

Another studio that I worked at was Secret Sound where I got to work with Spiro Gyra on Morning Dance, another one of my favorite records. I also got to work with Cissy Houston and also got to record Whitney Houston who at that time was still in grade school. She’d come to the studio in her school uniform and sing back up vocals with mommy. I also worked with Irene Cara and Luther Van Dross who was part of a background group called Kay Lee.

Then, in the early 80’s digital came along, and RCA Studios got heavily into it, more than any other studio in New York. Where was the most cutting edge digital happening? It was happening on 44^th Street and Sixth Avenue. I wanted that knowledge so I went and became a staff engineer at RCA. I spent a decade with them and made the transition from just a recording engineer to going into mastering. And, I made that decision when it was no longer vinyl that was being mastered.

We were using a JVC system because they preceded Sony in digital on tape. We were using a Soundstream system with Honeywell disc drum drives. The disc drums were about the size of a cake box and in them there would be maybe three 12” or 14” platters, stacked. And you would take these big drums and put them in what looked like top loading washing machines.

And then Sony came across with the 1630 system. Even though it’s an antique, the 1630 ¾” tape is still, to this day, the most flawless way to record a 2 channel digital and take it from Location A to Location B. And the way that we started doing mastering was to set up two 1630 systems, and then go from digital to analog. Then we would go through some of the wonderful analog gear that RCA owned, and then you’d end up on the 1630 system number 2. So you would willfully leave the digital domain and come back. And one of the curious things that we found was rather than the signal degrading, the number of errors got fewer and fewer. Three generations down sounded better than the original. Go figure. Many of the early CDs were made this way.

What is mastering and what makes it different from mixing?

That’s a good question, especially in digital audio. Mastering today is more of a science than an art. Part of it is art but if you asked me what part of it is art I’d say it is the artful application of the correct science. It has nothing to do with “This is how I think your record should sound.” Your record’s going to tell me how it needs to sound.

Is it necessary for the client to attend, or is it necessary for the producer to actually produce the mastering?

No, because it is the tracks that produce the mastering. The track tells me what it wants and if the track can’t tell me what it wants it’s not a good mix. And the corrections should not be made in the mastering studio. They should be made with someone who has access to the multi channel tapes. In this room I got two channels – I got Left and I got Right and that’s it. But you can still do an enormous amount with those 2 channels. What mastering is, is no matter how good the recording engineer is, there are going to be errors on that final mix. There are going to be technical errors on that digital mix. A lot of those errors today are from the systems that people use. The systems will impart errors because the systems are all pretending to be the Hit Factory in a box, or any of the great studios. You can’t have a million dollar SSL console and another half a million dollars worth of outboard in the virtual world. It’s a fairy tale. It does not exist. You can approximate it, and, while you are approximating you are introducing digital errors that you can neither see nor hear. Why?

Because after you’ve been listening to it day after day, week after week, something comes into play I call Sense Memory. Your Sense Memory goes back to the basic tracks of a recording. By the time you get to the final mix, the excitement that you are feeling is your Sense Memory of the tracks before all the overdubs. So, has your track lost its life? It may have. Will you know it? Absolutely not, unless you can distance yourself. And this is what separates the great mix engineers from everybody else. They can distance themselves in a matter of a couple of days. The average person is going to need 7 or 8 months before he realizes that the record he made 7 months ago is absolutely horrible although at the time he didn’t think that at all. I’ve had that reaction in my career as a recording engineer, “Damn! So that’s what that record sounded like? I thought it was better than that.” And the reason I thought it was better than that is because my sense memory was playing the track in my brain louder than what was coming into my ear.

So as a Mastering Engineer you want to limit yourself to no more than 6 hours a day. Beyond the sixth hour you can do some editing, and, maybe you can do a little leveling, but you can’t make any decision on how something is going to be EQ’d or limited or compressed because your sense memory is going to kick in. And it’s going to override the new information that’s coming in. It’s a human condition that we really can’t avoid.

I am now employing a radical approach to mastering. We used to master records in one day. And we used to come back the second day to see if any touchups were needed. And that would probably be an 8 or 9 hour first day, and, a 6 or 7 hour second day. And that would be the record. And that was doing everything the way we used to do it which was not using the computer but rather separate pieces of gear. The computer was only used for storage and editing. Back in the day when we were doing what I described earlier as what we used to do at RCA, from machine to machine, you had to do everything in real time. It was the only way to do it. As you come forward in time most of your gear doesn’t really exist but it’s the picture of your gear on your computer monitor. You aren’t really working with that gear. You’re working with a virtual simulation of that gear.

I started questioning that idea when I did my last installation at Media Force. We were still mastering the old fashioned way – everything will happen at the same time. Part of being a good mastering engineer is listening to raw tracks and deciding what the mastering chain will be. They’ll call it the path, they’ll call it the chain, but it is the whole idea of what is going to come first: the level control, followed by some limiting followed by some EQ, followed by some compression, followed by some more limiting, another level set and then we’re done. Now the mastering pathway changes with everything that you’re doing but that’s the basic idea. And that’s how most of my colleagues still work.

When you’re working with actual gear, and you’re coming from analog this still, to me, is the way to go. It definitely makes sense. You come from your tape, through your analog gear, figure out your mastering path, and the very last step on that path is make it digital and store it. However, come into the 21^st century and realize that a lot of studios are not studios anymore. They’re studios in a box where everything is being done virtually. And I questioned whether mastering could actually be done that way. And to me it made more sense that mastering could be done that way and that recording actually should not be done that way. Recording should be done at places like Troy Germano’s new place where you’re in a room that sounds just wonderful and the gear is absolutely first rate. And it’s all real gear that you can touch. And I think that’s how recording should be done because a lot of mistakes are made in the box.

But I came to think that mastering could be done in the box if you respect the limitations. The first limitation that I had to respect is that you cannot create a mastering chain. And the reason is that when you are mastering, the science is that you are trying to create a flawless digital signal. Whenever you do any compound procedure on a computer you are introducing new inaccuracies. You get the bad with the good. There is no purist; oh this is only giving me the good, That’s a fairy tale. You don’t get something for nothing. Every time you get something from digital you’re giving something up. Because you know you have that finite wall which dictates that there should be nothing above digital zero because beyond digital zero there’s going to be a minor ticky clicky thing. And it will happen so fast that you’re not actually going to hear it unless you have a lot of them happening at one time and then you are going to hear it. Or, you’re going to hear, “Hey, your music sounds digital.” That’s because you’re probably going over level 15,000 times in a three minute period. But it’s so brief a moment in time your ear does not identify it as distortion but overall your psyche says, “There’s something wrong with that recording. It doesn’t sound friendly. Yeah I think it sounds good but there’s something about it I don’t like.”

While I was mastering at Media Force I realized that a lot of the stuff coming in sounded really great but only to a degree. What is wrong? As the analytic tools got better you could actually get an error count or you could get a count of how many times the thing went over zero, or, is there a DC offset; meaning that, is absolute null, no sound whatsoever, where it needs to be?

When you look at a waveform, the waveform goes above and below the null point. But that null point sometimes gets askew. And that is caused by making compound procedures on any computer. Very often I will take in a mix and the first thing I’ll look for is, is the null point really null? Is nothing really nothing? Or is nothing really above or below the null point? That affects not only the silence between the notes but also the resolution and decay of every single sound on that record. If you straighten that out you might find out that the mix only needs one fiftieth of the EQ you thought it would. And now, the bass sounds more like a bass.

I started talking to a number of guys who actually created this virtual gear and one guy told me, “ If we didn’t have to waste so much computing power and time on making a front end that made you guys happy we could accomplish so much more with the program doing so much less.” What he meant was that most engineers want to see their virtual 1176 looking like an 1176. They want to see the knobs; they want to see the meters. The 1176 really doesn’t exist, it’s a picture of it. Meanwhile, when you adjust the threshold of that 1176, or the ratio of compression on that piece of gear, in the virtual world you’re not actually doing that. As the programmer told me, “When we give you that program it doesn’t mean that your computer has turned into an LA2A or an 1176. It is the simulation of what that device would do if you were using it.” So what they had to program into all of these things, especially into Pro Tools, is something that looks like the gear that you want to touch. And every time you touch a knob on that gear, what’s happening behind the scenes are things that most engineers don’t understand.

I began working under my new radical theory, which goes to what I was saying before about what happens when you get a computer system to become a virtual limiter. Well, yes it makes some errors because it has to translate from the front end, which is make believe. And it’s just to make the engineer who has been turning knobs and pushing buttons feel comfortable. Not only are there mistakes that happen in the computer because of that but now the big thing of compound errors. And remember we’re talking about computers with one or two CPUs and as powerful as they are, they have their limitations. When you start doing compound things like I’m going to use its virtual limiter, its virtual equalizer, both plugged into this virtual fader, well now you’re asking the CPU to do 3 things. And in your engineer’s mind you’re seeing them the way they would be if they were separate boxes in your room. And that would be you’d go to whichever one you decided would be first in the chain, and then the next, and then the next. And any engineer will tell you that even with the same settings, if you change the appearance of who’s first, who’s second, who’s third, the result is going to be vastly different. In the world of the computer, the computer is trying to let you know that it’s not going to let the results be that different. Within that CPU, all three of those devices, let’s just pick a level control, a compressor, an EQ; all three of them are going to use common computing procedures. So even though you change device number 2, it’s actually changing the performance of device number 1.

Now in the real world this can never happen because what you’re doing on device number one is still going to work the way you want it to. Then you’re going to go into the EQ. But no matter what you do with the EQ it’s not going to reach back and change the compression ratio. In a computer, it is. The way you work around that is to do one thing at a time. This goes back to what I said earlier that recording should be done in a proper studio with proper gear. Because you don’t have the luxury of doing one process at a time in a multi-track situation. However, as a mastering engineer you do have the luxury of doing one process at a time.

What kind of computer system do you have?

I’m actually working on a PC. I’ve found that in the year 2009 the PC, with Windows XP is very friendly.

What about overall level?

That’s been a bone of contention not only with me but any top-notch guy you talk to. The need for the CD to sound louder is the silliest thing. I understand that you want your CD to command presence but to pursue loudness for it’s own sake is just silly. Because for every db of loudness that you add what you start to give up is the resolution of every single note. You don’t give up much of the dynamic range. However, every note has a resolution decay and you start giving that up when you start pushing beyond. There is a scale called the Root Mean Scale, RMS. It is a minus scale which can be used to measure perceived loudness.

However, I can show you a recording that is so soft that you have to turn your amplifier all the way up to hear it, and yet, it reaches digital 0. It reaches the top of the digital scale but when you measure the analog resolution it’s almost nothing. A very handy piece of gear is a precision set of VU meters. I’m not talking about program peeking meters. But if I want to look at how I can relate to this RMS level I have to look at a standard VU meter. That shows the analog resolution that I’m getting from my digital. What I tell most mixers is not to rely on their digital meters. They’re too fast and they’re even going to see false peak information.

What should someone look for in a mastering engineer, assuming you’re not available?

I’d have to say how long has he been doing it? That to me means a lot. Did he start out as a mastering engineer or did he start out as a recording engineer. Most digital mastering engineers started out as recording engineers. Exceptions to that rule: Bernie Grundman and Bob Ludwig among others. They started out as vinyl mastering engineers. So, unless you started out as a vinyl-mastering guy you better have started out as a recording guy. I’m going to look at how long you work in a day. How much work are you willing to take on. Do you believe that you can be a victim to A, fatigue, B, the Sense Memory? Another thing I would look for is, if you’re artsy, I don’t need you. If your art is which science to use, you’re my guy.

Any final words of wisdom?

Yes. Buy yourself a decent set of VU meters and for every element that you add to your multi-track, while you’re tracking it, make sure your eyes are glued to that VU meter. And make sure that you’re not at the top of your digital scale on your digital meter, lighting the little red lights. And, if you need to, turn your monitors up to kill levels, fool yourself so you’ll be forced to just barely crack the faders open and then look at your meter.