If it’s true that “a little knowledge is a dangerous thing”, then that adage probably applies double to the art of music mastering, where adherence to some dubious practices has had a widely discussed negative effect on the quality and musicality of recordings over the last couple of decades. With the widespread availability of mastering tools that can be easily applied in any DAW, even relatively inexperienced mixers are likely to also take on that task, but professional mastering engineers know that it’s a distinctly different art from mixing, and one that’s all about careful listening and building up a wealth of experience with many different mixes, not just one’s own familiar efforts. With all that in mind, here are a few areas where many people are likely to have a basic working knowledge of how to approach a particular mastering task, but not enough experience to get the best results they could.
Everybody knows that mastering nowadays involves applying overall processing to a stereo mix, something that’s easy to do with the wide availability of mastering tools for DAWs. In the old days, mastering was always a separate stage of the production process, applied to the finished mix by a dedicated mastering engineer. But nowadays many mixers combine mixing and mastering, strapping mastering-type processing across the master stereo output of the full mix session. Some even go so far as to mix through that processing, on the theory that if it’s going to be eventually applied anyway, mix decisions may as well factor it in from the get-go.
Of course this can be a perfectly good approach—after all, if it works, it’s all good. But others, myself included, tend to feel that mastering and stereo bus processing in general may be better as a separate, final stage: the more traditional approach. As many mastering engineers would suggest, ideally a mix should be able to stand on its own, without heavy stereo bus processing, and the processing applied at the mastering stage should be geared to small, subtle adjustments, intended to take something that already works as is and make it work just a little better.
And there’s some benefit to having a fresh set of ears listen to the mix; again, this harks back to the old days, when mastering was more about dealing with purely technical issues than trying to somehow make the mix “better”. In fact, a good mix may need little or no processing at the mastering stage, and there’d be nothing wrong with that.
Fixing The Mix?
That brings up another point about mastering. Many mixers look to mastering primarily as a way to “improve” the mix. There used to be an old saying when problems arose during recording, “we’ll fix it in the mix”; these days that’s shifted to “we’ll fix it in mastering”—the idea that problems with a mix can be deferred to the mastering stage to be addressed. But the primary goal of mastering shouldn’t really be to make the mix sound “better”; ideally it shouldn’t necessarily change the mix at all. Mastering processing can be as simple as slightly adjusting overall stereo balance, correcting any inter-channel phase issues, or simply matching overall loudness for a collection of songs.
If mastering is used to make subtle improvements to a mix, these should be very small tonal tweaks. Like a slight increase in brightness or bass (sometimes as little as 1 dB is enough) to promote slightly greater clarity or fullness, or to add a little compression just to maintain a more consistent overall dynamic, when appropriate. A good reason for small tweaks like this might be to ensure that the overall tonal balance and dynamic range of the mix falls within current norms, so it doesn’t attract negative attention when heard as part of a playlist or stream, but not necessarily to make the mix sound “bigger” than it is.
These kinds of subtle considerations are really what mastering should be focused on, not trying to take a mix with problems—that, say, suffers from uneven dynamics, or has specific instrumental imbalances—and fix those problems at the final stage. While those kind of fixes can be and often are done as part of mastering, ideally such issues would be best addressed back at the mixing stage. It’s not unheard of for a mastering engineer to suggest that certain problems would be better remedied by the mixer, with the adjusted mix then sent back to the mastering house for finalizing.
The Loudness War
No discussion of mastering would be complete without a nod to the infamous Loudness War, where every song is brickwall-limited and has its average loudness level cranked up to be as loud as possible. It’s used to draw attention to itself, though at the expense of musical dynamics, punch, and clarity. Even novice engineers quickly learn that if they don’t apply heavy limiting—a.k.a. Maximizing—and push the level way up, their lower-level songs run the risk of being a letdown when heard as part of a playlist or stream of louder mixes, and this can subconsciously result in a negative reaction to the music itself. But that well-established practice is finally changing.
Recent music distribution channels—CDs and digital downloads—never offered any way to prevent the excessive use of loudness maximizing to ensure reasonable average levels that would allow for less-compressed, better-sounding masters. Mandated loudness standards have been in place in the broadcast area (tv, film) for years now, eliminating scourges like loud annoying commercials, and with the rise of streaming as a primary source for music consumption, similar control is now trending for music distribution as well.
The largest music streaming services now mandate lower, more reasonable average loudness levels for music heard in their streams; DAWs are now including loudness metering to augment/replace the traditional peak metering, and the trend is toward lower average levels, which is looking like it may be bringing the Loudness War to an end. So the knee-jerk practice of maximizing level in mastering should be considered old hat, maybe applied only to masters for specific release formats (CDs?), with most masters finalized at more reasonable levels. There are many articles addressing the specifics of this trend, and anyone involved in mastering at any level should become well-versed in the relevant technical issues.
As I mentioned up top, traditional mastering was applied as a final stage to the finished stereo (originally mono) master. Originally, it was mainly to prep the finished mix for vinyl, which had a more limited dynamic range than tape, and address other potential playback issues which required specific technical tweaks. But the goal was always to be as true to the sound of the finished mix as possible. Even when multitrack recording took hold, this was still the case for mastering.
Nowadays, while it’s still most typical for the mixer to provide the mastering engineer with a finished stereo audio file, some mixers have learned that they can provide the mastering guy with stems as well. Stems are submixes of the main instrumental groups—for example, all drum tracks, all bass tracks, all guitars, all keyboards, all backing vocals, lead vocals, etc. The stems would include all the same processing as they’d have in the full mix, and playing them all at Unity Gain should be identical to a stereo mix of the same session. The idea is that this can provide the mastering engineer with more flexibility than he’d have with only the finished stereo mix, allowing him to make more significant adjustments to the balance of those musical elements without the need for special processing techniques (like MS).
But this brings up the issue of whether those kinds of musical adjustments should really be part of a mastering session at all. Now of course, if the same person is handling both mixing and mastering than the issue is moot—he’ll always have access to the individual mix elements and could go back and make small tweaks to the mix at any time. But when sending a mix to a dedicated mastering house, should stems be included as well? Should the mastering engineer even be expected to make those kind of adjustments?
Some mastering engineers may welcome the option of stems, especially if they’re doing a mastering job for a client whose mixes often have flaws that need greater adjustments than mastering techniques usually afford, and where studio politics preclude sending the mix back for the fixes. But in general I tend to think that the kinds of tweaks that would require stems really should be addressed back at the mix stage, by the mix engineer. It’s normal for the mastering guy to use special techniques like MS (Mid-Side) processing to slightly adjust things like lead vocal level against the track, but many mastering engineers feel that more intensive changes to the mix should really be referred back to the mixer.
What is good practice is to provide the mastering house with a couple of alternative versions of the mix; traditionally, this meant vocal-up and vocal-down versions in addition to the primary mix. These would have the lead vocal about 1 dB louder and 1 dB softer, with everything else the same. The idea is that the balance of that critical element can often sound different on different systems, and if the mixer’s monitoring environment is a little midrange-heavy or midrange-shy, then the vocal-up or down version may be in better balance when heard over the mastering house’s assumedly more neutral reference monitors in their more carefully optimized environment.
Nowadays there’s another possible consideration when it comes to alternates, relating to stereo bus processing. Most mastering engineers prefer the the mix have little or no stereo bus processing—this provides them with greater flexibility to do whatever they need to do to achieve optimal levels and overall dynamics and tone. If the mixer over-processes the mix, then the mastering engineer has less control to successfully make any small adjustments he deems necessary.
Some mixers like to apply relatively heavy buss processing to their mixes, and consider it part of their method. This is fine, but I’d still suggest making a bounce without that processing in addition to the ‘preferred’ one with it. That way, in a worst-case scenario where the mastering engineers feels he can’t make a problematic more-heavily-processed version work technically he can turn to the less-processed alternate, and use the processed version as a guide to what the mixer was shooting for—he may be able to achieve that sound without any concurrent technical problems. If the trend toward lower-level, less-processed mixes continues, a mix that can’t stand on its own without very heavy bus compression and EQ may be at a disadvantage.
For those handling their own mastering, it might be a good idea to create a couple of different finished masters—one at the ideal average loudness level (i.e. -14dBfs/LUFS, optimized for streaming); one with a lower average loudness (-20 or less, for future-proofing), and maybe one that’s maximized (up to but not exceeding -9dBfs/LUFS, for older media like CDs).
Mastering is an art unto itself, and while many standard practices are well-established, others are changing, so it’s worth a little extra effort to get familiar with current trends and alternative approaches. Ultimately, the most important thing is that a finished master sounds good against the best-sounding commercial recordings, whatever methods are used to achieve that end.