However, not everybody really knows what to listen for.  It can be like asking someone wearing dark sunglasses what they think about the color of the curtains – they’ve got a vague idea of what’s going on but aren’t really sure what they’re looking at.

Overall, it’s pretty subjective.  There are a few objective criteria but most mastering engineers worth their salt will get that right regardless.  Additionally, a mastering job can often be very subtle, so if you’re not really listening for something very specific, the masterer’s work might not sound terribly obvious.

Still, on a broader scope, there are things to listen for that any musician can grasp pretty intuitively:

• Is it consistent?  Any good mastering engineer strives to make every song on an album sound like it belongs with every other song, whether that’s by loudness or EQ or overall feel.  Sometimes, though, consistency is in the eye of the beholder.  A compilation, for example, or even an artist that switches styles a lot, can provide a challenge – what the artist (or compiler) thinks is consistent might not be the same as what the engineer (who may not have any familiarity with the genres or styles) thinks is consistent
• How are the dynamics? Despite the whole issue of the Loudness War and the pros and cons of compression and limiting, it’s still pretty much a given that anything handed to an engineer is going to get at least a little compression or limiting to level things out and change the overall perceived loudness of a track.  The overall extent of this, though, is generally left to the artists to decide, though.  I can easily take a track up to a mean RMS of +10db (K-14) – that’s very loud.  Most artists may not want that.  I can keep it at a mean RMS of +5db (K-14) with a very high crest factor – that’s quiet and very dynamic.  A lot of artists won’t want that either.  Some artists, particularly in dance, WILL want a high mean RMS at the expense of dynamic range.  Jazz artists might very well want a very low RMS and high dynamics, even at the cost of some pleasing “glue.”  A lot of it depends on the song too.  Do you want it to sound open and airy or dense and punchy, or something else entirely?  A mastering engineer will have opinions and generally a pretty good sense of where to go on this, just based on experience, but it can still be intensely personal (or even dependent on application).
• Does it match your goal for the sound?  Since digital signal processing has made a lot more “sounds” possible than in the past, a lot of mastering engineers have a wider palette to work with – without being tied to just one hardwired signal chain, they can re-route and emulate as need be (although not all of them do.  Abbey Road, for example, has a killer mastering service and they use the same sort of bespoke gear they’ve been using for decades.  But they have a “sound” all their own too).   As such, the result may not actually be what you’re looking for.  I once had a mastering job that I found to be rather harsh and cold, so I gave it a very warm analog sound, as I thought fit the material – but the client came back to me and said “well, that’s really nice, but I was looking for this to sound abrasive and noisy.”  In short, my polishing of the material was exactly the opposite of what the client wanted, and their harsh, digital sound was intentional and not just the result of their gear.  This comes down to intuition and opinion – the engineer may do what he thinks is right for the material, but that may not be what the artist intended.  Some of this can be alleviated before the process even starts, with communication of the end goal, but as is often the case with art, these things aren’t always obvious.

For the most part, what you as a client will be evaluating from your mastering engineer will be simply be how much what you get back lines up with your eventual vision.  The engineer will have to make decisions based on how well the mix will translate across sound systems, how it will stack up on the eventual release with other tracks, and other such choices that could affect how you want the track to be perceived.  Usually, it’s a pretty transparent process of just putting the final shine on the whole album, but in the end, as the client, you have the final say.

Never actually owning a G-Series console, I can’t vouch for the accuracy, but comparing it to my other VCA-modelling compressors, even the 6030, it’s got a “smackiness” and a thick low-end that I can’t achieve most other ways.  It’s stellar strapped across the 2bus of a rock song.

The original G-Series lacked a few of the nice features of The Glue – it doesn’t have a “range” knob to limit the amount of gain reduction, and it didn’t have a peak limiter, and it didn’t have a superfast attack setting.  In that regard, the Glue loses points on accuracy, but gains them on usability.  The peak limiter is a little iffy – it does a decent job, but it’s not a subtle effect, and if you’re using the peak limiter a lot, chances are you’re hitting your 2bus too hard.  The Glue does model the G-series fixed ratios -1, 2,4, and 10x and the stepped attack and decay, so in that regard there’s very high similarity to the original gear’s workings.

It does sound great, particularly on aggressive material.  However, multiple instances of The Glue on multiple tracks or busses can be a bit wearying – if the Glue is used on a drum bus and also on a 2bus, the end result is going to be fatiguing.  Additionally, the interface for selecting presets is a little weird – they’re not preloaded into a menu, you need to click on a box to get a dialog where you can select a preset to load, and…well, it’s a bit cumbersome.  That said, it’s a compressor with 6 knobs and two switches, so if you’re desperate for presets, you’re likely to get more benefit by reading up on compressors, since this plugin seems to be targeted towards the finicky tweaker type.

At $50, too, I guess it became pretty popular, because Chris Johnson recently pushed out BussColors2, which adds a lot more power and functionality to the plugin.

The first Busscolors was basically a pass-through that ran the incoming signal through one of four convolution curves – “Rock” (SSL), “Lush” (Neve), “Punch” (API) and “Tube” (um, something tube-y).  The curves weren’t exact point-by-point samplings, but were a few samplings smoothly interpolated, which is really good enough for most.  Tube added nice airiness, Lush smoothed things out and gave a little more low-end beef, Punch was, in fact, punchy, and Rock had the 80’s-SSL smackiness.  The thing was, they were incredibly subtle, and to get much benefit from it you had to use multiple instances.

Busscolors2 adds a number of new models – most notable “Dark”, which models an MCI console.  Some of the other new models, like “Vibe”, “Steel” and “Holo” don’t model anything directly but seem to be designed to act like consoles without sounding like a specific one, instead designed to solve problems.  “Vibe” is somewhat Neve-y, but without the low-end beef (which can build up across instances).  “Steel” thins out the low-mids, allowing for more clarity in bass-heavy electronic compositions.  And so forth.

The additional models would be a stellar upgrade (the MCI, particularly, sounds great across a drum bus).  But going one better, input and output gain controls have been added, and somewhere deep in the bowels of the plugin, a saturator has been built.  So not only can you drive it hard, it will eat peaks appropriately.  Subtly used, it can shape a track or a bus nicely.  Used more aggressively it can sound like actual gear that’s being driven hard to distortion.

Most of the time it’s to be used subtly.  I find myself using the MCI console across drum busses a lot, with a very slight input gain and a very slight reduction in output.  It does a nice job of smoothing the rough edges of a track without brutally muddying things up.

This has led to plugins that do nothing more than model the slew rate behavior of a console,  or plugins that emulate certain analog summing functions, or plugins that do nothing but digital signal dithering.  It can tend to be esoteric, but it can also tend to be kind of stunningly useful.

ToTape (and the free companion “FromTape”) is one in a line of his tape saturation modeling plugins – previously including the popular “IronOxide” saturator.  What’s particularly interesting about ToTape is that, rather than actually modeling tape itself, it models the behavior of the tape.  Consequently you get the saturation and “glue” of the tape without the coloration.    This makes it very flexible, although definitely not an all-purpose plugin – sometimes the coloration is exactly what you’re seeking to emulate.

The FromTape companion plugin is a subset of the ToTape controlset, intended for use on tracks and busses.  It models just the “compression” part of tape saturation, without any of the equalization or leveling.

The aforementioned EQ and leveling parts are the more interesting parts of ToTape.  While some tape emulators, like McDSP’s AC202 and The UAD A800, allow you to vary tape speed, bias, and all those actual tape parameters to achieve certain sounds, ToTape merely gives you “bump” and “treble soften.”  Whereas most tape emulators give you a fixed bump and rolloff to emulate a certain type of tape head, and treble effects are achieved using bias, tape speed, EQ type, etc, here you’re given a pair of sliders.  This gives you very fine control over the output effect – you can pair a low bump with a high “treble soften” in a way you couldn’t achieve with tape.  Of course, the output then doesn’t really sound like a tape, but it may still give you the sort of saturation and punch you’re seeking to achieve.  The actual “tape slam” is akin to a combination of input drive and variable tape compression, which again can be driven in a decidedly non-tape-y way, and it even has the capacity to overdrive the output signal, which tape can’t physically do (there is, however, an output trim slider for such an occasion).

It’s not an all-purpose plugin, and is probably best used judiciously, but it can do some wonders to a track in need of a little warmth or saturation.

Or should I say “suite of compressors.”

The value of McDSP 6030 isn’t in its full-featured-ness as a compressor, it’s in the fact that it’s basically 10 simpler compressors, each with different characteristics, that can be switched seamlessly.  There’s everything from a simple one-knob, LA-2A style opto compressor to a feed-forward dbx160 type compressor, making it incredibly versatile.

Of course this raises the question “so, okay, it’s versatile, but how does it sound?”

A valid question, since there are a few multi-tool compressors and EQs that are only sort of okay at what they do.  But in a word, 6030 sounds great.  It’s warm, punchy, and some of the models can be driven extremely hard. The McDSP-designed “FRG444” can squash a signal down to a mere suggestion of the original dynamics (although to be fair I’ve not yet found an actual use for this one).

The collection of compressors includes emulations of a Fairchild670, a Neve 33609, an LA-2A, a Mu-Tube, and then…well, to be honest none of them are exact models, but that’s not a bad thing.  There’s a vaguely dbx160ish one called “Over-EZ”, something kind of SSL-esque called “SST76”, a compressor modeled on the Empirical Distressor called “D357”, an one mystifyingly called “iComp” which is blue, seems to be an intelligent-ratio kind of thing but is basically an all-original creation.  None of these compressors is dead-on – the U670 for example has slightly different ballistics and doesn’t have the Lat/Vert controls of an actual Fairchild 670.  The Opto-C deosn’t sound exactly like an LA-2A.  But they all “feel”, for lack of a better word, like high quality compressors of that design.  U670 feels like a vintage tube-based compressor, British-C feels like a Neve – the response from the compressor is what you’d expect from those pieces of hardware, even if they’re not dead-on impersonations.  Whether this is good or bad is a matter of interpretation – for me, it’s good, since I’m not too concerned with sounding exactly like I have a vintage Fairchild in my basement, merely that I have a good, smooth, tube-y compressor for vocals.

As both a blessing and a curse, the makeup gain stage of 6030 saturates instead of straight-up clipping, so once you’ve compressed the signal you can crank the makeup gain well beyond what would normally be a digital over.  The saturation algorithm used is generally pretty smooth but overuse on many tracks can of course make things muddy.

The price of McDSP 6030 (native – TDM/RTAS costs more) is $229, which for an individual compressor plugin is a bit on the higher end side.  However, for that price you get a very flexible compressor toolbox that outclasses a lot of plugins in the same pricerange.

The Solution: The track itself was a quiet piano-and voice version of one of the main album tracks, with high dynamic range and a startlingly good performance.  However, it was recorded separately from the rest of the album with lesser gear and little production, and thus didn’t match the wide, beefy bombast of the rest of the album.   There was also an extra bit of noise at the end that needed some deft editing to remove.

Careful application of EQ was applied to surgically bring out the vocals and get a little separation from the piano.  From there, some M/S processing on separate EQ bands was used to move the piano more out to the sides and center the vocals.

Editing out the etxra bits at the end was difficult – a simple fade would’ve also cut out the actual tail of the piano and sounded unnatural, and a straight-ahead cut-edit would’ve sounded even worse.  Instead, a cut edit was applied, but the final tail of the piano to the point where the noise began was timestretched very slightly and faded, to give it a more natural tail, while some “silent” room ambience sampled from the beginning of the track was crossfaded in to provide more natural-sounding ambience.

With the exception of the very-high-end stuff, like SaDiE, most attempt to integrate with an existing DAW and focus solely on the audio side of mastering and ignore the media-output end of things. But that’s not really why I don’t like them.

Now, to be fair, most of them do their best to provide some fairly impressive functionality. IK Multimedia’s T-Racks suite, for example, has some very lovely options for emulating a Pulteq EQ and a Fairchild 670 compressor. iZotope’s Ozone has some very powerful options for mid-side processing on material. The problem is that these packages inhabit a very strange no-man’s land in the market. They’re more expensive than the software that comes with most DAWs and offer more functionality in many ways, but at the same time, the kind of engineer who is going to need that sort of functionality is going to likely want even finer control and higher-quality than these intro-to-midrange packages can provide.

It’s sort of the Swiss Army Knife problem. Swiss army knives are really cool. Some of them have a zillion little tools, from tiny screwdrivers to folding pliers to scissors. They’re great in a pinch, although some of the fancier ones are rather pricy. That’s just it, though – they’re great in a pinch, but you wouldn’t want to do an awful lot of real repair with them, and for the price of the one with all the fancy little clever tools on it, you could buy a pretty decent kit of basic full-size tools, a socket set, and a respectable power screwdriver. All-in-one suites are pretty neat, but like Swiss army knives, why spend the money on such a fancy one when you can get a bunch of “full sized” tools for the same money?

Perhaps that’s a little hard on some of these packages, as they do often have “full-sized” tools in them. The problem is, as I see it, that not everything in one of these suites is going to be perfect, you’re often paying for a lot of functionality you’re not going to need or want, and quite often you get stuck with a characteristic sound over which you have little control. Take, for example, T-Racks. As I mentioned, the Pulteq EQ and Fairchild Compressor emulator is pretty nice, and certainly for the price it out-competes an awful lot of stuff in the same price class. But some of the other modules are not so great, and you may find that even the really nice stuff isn’t what you need in the mastering stage (but would be more appropriate for tracking and mixing). So you spend $500 MSRP for a bundle with a great compressor and EQ that you never use for mastering, and some middling other plugins that you do. Similarly with Ozone, there is certainly some power under the hood, particularly with the M/S processing. Frankly, though, the overall sound is decent but not pants-wettingly spectacular, and a few of the inclusions seem a bit baffling to me – the “mastering reverb” for example, seems like a module of extremely limited use. If you’re at the mastering stage and you find you need to add reverb, something has gone wrong. It’s not a bad-sounding reverb, but it’s not a great one either. We’re also back to the “what market is this for?” problem – if you’re mastering your own material, you can go back and add your own reverb to the mix with much more precision than any full-track mastering ‘verb could; if you’re mastering this for someone else, there’s a good chance you have already invested in better gear than this and could apply it yourself.

Then there’s the preset problem. I’ve had the opportunity to bash my way through some of these suites and they all seem to come with a pretty deep library of presets. That’s all well-and good, but even in my limited experience as a mastering engineer I can tell you that no two tracks will ever need the exact same settings. Sure, presets always do make a great starting point, saving you a little time dialing in some common settings, but the urge to just fire-and-forget, especially in the neophyte stages of the process, is strong. Some of the presets in any one of the packages have the terrifying ability to sound completely awesome, but only in the context of a good set of speakers. Hit Ozone’s “enhance and widen” preset and you’d get a lush, wide, open mix on a good set of monitors…that would sound washed-out and phasy on a club system. Such a preset might help rejuvenate an old mono mix or something pulled off an archival 4-track, but it has the capacity to completely screw up a modern recording. Through the joys of psychoacoustics and comparative listening, after hearing that, every other track will sound dense and narrow and feel like it needs the same effects, which will mean an *entire album* will sound wide and airy on the monitors and then washed and phasy on a club, or ear-hurtingly trebly on a car stereo, or grainy and weird on an ipod.

The kitchen-sink approach is also a little worrisome. I have come across very few mastering jobs that require not merely the same settings, but even the same effects chain. Some won’t require any extra compression. Some won’t need EQ. Some won’t need any maximizing/limiting. So there’s no need to even have these effects patched in, much less turned on. When you have a suite with 8 or 9 effects available at all times, you have to know and know well what you need and what you don’t, because turning on a multiband limiter when you don’t need one is going to dramatically change the sound of the output. Certainly, messing about with stereo width is always a very dodgy proposition, because the capacity for phase problems is extremely high, so many mastering chains don’t even bother (or, if they do, use some M/S processing for the effect) – so having a dedicated stereo imaging processor available at all times is sort of the mastering equivalent to keeping a loaded gun on the nightstand: you may never use it, but just having it there is risky.

This is not to say these things are not without their place. I’ve mentioned that Ozone’s M/S processing is really excellent. The problem for me is that I don’t necessarily need to buy an entire suite to just do M/S EQ or compression, when I already have the ability, albeit with a little extra bus routing, to do M/S processing using any plugin or hardware I like using only the stuff that comes with Logic. Sure, it might be easier to just hit the MS button in Ozone, but because I now know how to accomplish this from first principles, I can get even more precise, and do multiband M/S processing if I really desired, or use M/S processing with anything, not just the processors that have it enabled. Or I could do it as simply as just increasing the relative levels of the mid or the side without processing either one. I didn’t need to spend $400 to keep myself from having to learn a useful skill.

And there, right there, is the crux of it. Mastering isn’t about the gear as much as it is about the ears and the grey mushy thing between them. Don’t get me wrong, great gear is awesome and makes things sound even better, but knowing the deep science of the process and knowing exactly what to listen for is the real trick, and no mastering suite software is going to help that.

While maybe it’s true that the mastering engineer *can* indeed fix these things, the smartest policy is to fix these things in the mixing stage first. Any fix in the mastering stage comes with a cost, whether that cost is monetary (extra engineering time means extra $$$) or just the cost to the overall quality of the track. That kick drum too boxy in the 200hz range? The mastering engineer could do a very precise reduction on that frequency…but anything else with harmonics at that frequency would also be affected. Vocals need to be louder? A little middle-band compression may help tighten that up, but at the expense of the overall width of the track. It may be subtle, but it’s going to hurt.

I’ve often heard of mastering referred to as a “surgical” process. That may be true to a point, because the tools are usually precise, specific, and often dangerous for the unpracticed to play around with. The analogy sort of breaks down at that point.
A better analogy would be perhaps one of a car detailer. The guy’s got special tools – an orbital buffer, polishes, hammers for pounding out dings, stuff that an average car owner probably doesn’t keep handy. The detailer’s job is to make your car look as good as it possibly can, buff it to a high gloss, maybe smooth out some scratches and chips. But you wouldn’t ask the guy who details your car to fix a problem with your head gasket or your muffler. The mastering engineer is like the detailer – he’s got the tools to make your track shiny, polished, and professional-sounding. And while he may have the ability just by chance to fix a mix problem (much as the detailer probably knows how to change your oil) it’s not the sort of thing you really want him to be doing.

It’ll save you, the artist or producer, time, money and headaches to address any mix problems at the mixing stage, and not the mastering stage.

So what about the case where the engineer hears a problem you missed? A good mastering engineer would likely tell you about it. In the world of the radio-hit machine, maybe not, because it has to be done yesterday and ready for the radio, but there it doesn’t come up as much to begin with because your track is mixed by Mark Stent or Tony Maserati and it’s just not an issue. For the rest of the indie pro-sumer crowd, though, the option is usually there to send the engineer a new mix with the fixes in place, and any resulting charge will be minimal, if there’s any charge at all. I know that when I master an album, I listen to it first so I know what I’m going to do before diving in, and I’ve on more than one occasion said “hey, the bass is a little loud on track 3” or “that lead synth has some weird phasing, is that intentional?” and if it’s a problem and not an artistic decision, I’ll gladly take delivery of a new mixdown before I start mastering (there are limits, of course – eventually I’ll want to get to mastering and not waiting for the 8th upload of a new mix, and sometimes the engineer does have the right to say “look, I don’t know how to fix it or I just don’t want to – do what you can with it.” But these are rare cases).

In short, the strongest recommendation I can make to anyone about to submit their music for mastering is “give them the best mix you can, and don’t expect the mastering engineer to fix what you didn’t.” A strong mix will need less processing in the mastering stage, leading to a cleaner, better-sounding master and a better experience for the listener.