If you’ve read my article “Mixing with the Professionals”, you should have a very decent basis on which to build more in-depth audio knowledge. In my experience, a lot of people look at the process of mastering as an easier task than mixing. The reality is that mastering can be just as complex and intricate as mixing, and even more so in some regards.
The following is an expert post by Jim Emerson. Jim holds the title “Executive Audio Engineer” at Reasonzone Audio. With a deep-rooted passion for his work, he has had the opportunity to help musicians around the world realize their creative potential.
Common Mastering Mistake
A common mistake I see is that people deem mastering somewhat of an afterthought: “I’ll just throw a limiter on there after my mixdown and crank it up to the right volume.” Unfortunately, mastering isn't just about volume, nor is it confined to compression and limiting. It’s about finding the perfect proportion of loudness and dynamics, perfectly attenuating the frequency curve of a song, perfectly calibrating it for the widest range of speakers possible, and most importantly, perfectly bringing out the nuances of each individual song and guiding it towards its fullest sonic potential.
People/guides/blogs are going to try and give you values to follow or methods to reproduce, but the truth is this: there isn't a surefire way to master a track. It requires a seasoned set of ears and time. I’ll ask you to keep this in mind as we trudge onward: I believe the best way to teach someone how to master is to walk through the process of mastering a song. In tune with my article on mixing, we can break the mastering process into manageable chunks:
- Chapter 1: Organization of a mastering session
- Chapter 2: Enhancing what’s there
- Chapter 3: Reducing the clutter
- Chapter 4: Master bus processing
- Chapter 5: Automation of master bus elements
Hopefully you, the reader, will gain some insight about what goes on during a professional mastering session, and can learn valuable techniques along the way.
Chapter 1: Organization of a mastering session
As with the process of mixing a track, it’s important to ensure that your mastering session is organized before beginning. However, unlike mixing, the organization process isn’t as time-consuming and rigorous when it comes time to master your file.
The very first thing you should do, as common sense would have it, is to ensure that your file is the correct version of your song and isn’t limited or compressed already. I know it seems like a given, but you’d be surprised how many files I get sent from clients that end up needing to be replaced. It’s fine and it happens, but this step can save you some headache later on, as your master chain will be different for every song you work with: you may end up having to completely redo your chain. Furthermore, it’s important to make sure your file wasn’t clipping when you exported it. This will result in loss of information that would otherwise be there and you’ll get a master that isn’t as good as it could be.
Most digital audio workstations will have some sort of “stretch” feature which serves to time-stretch audio and sometimes MIDI clips. It’s important to make sure that this is turned off for the clip you import to master. You can get some weird artifacts if it’s left enabled, and your song can come out at a different tempo than the tempo you intended it to be.
Chapter 2: Enhancing what’s there
The very first set of actions I take after organizing my mastering session actually doesn’t take place on my master channel. I find it very useful to bus the non-mastered mix to a pre-master channel for some touch-up and doctoring. You could do everything on the master channel, but I find that keeping the stages of my mastering approach separate results in an efficient session and tasteful end product. When I “enhance what’s there” before my master processing, I find myself lumping everything into a few different categories:
- Enhancing pre-master EQ
- Enhancing the stereo field
- Adding “sheen”
Enhancing pre-master EQ
Something I find myself doing on a lot of my clients’ tracks is enhancing the pre-master EQ curve. A lot of engineers will tell you that this is unorthodox and that you should leave the spectral balance as the mixing engineer intended, but increasingly common is the fact that people are mixing their own music at home. More often than not, files that we as mastering engineers receive are self-mixed in an environment that isn’t acoustically treated and can use a bit of touch-up as far as EQ is concerned.
Throwing back to my article on mixing, a well-balanced track will have a sonic spectrum that resembles the balance of pink noise. Of course, there are going to be exceptions and there are going to be mixes that are intentionally skewed to be heavy on the low-end, etc. However, in almost every mix I’ve received to master, some broad, subtle EQ boosts and cuts across the spectrum are welcome and can really bring out the good in a track.
Obviously, if you are mastering your own mix, it’s probably better to go back into your mix session and make necessary changes there rather than during your mastering session. It’s a good practice to get things right in the mix before mastering, and I think we can all agree on that. It’s just my philosophy that sometimes it’s fine to forego this approach and let your EQ’s do their job.
Enhancing the stereo field
As with EQ enhancement, some may say that the stereo field is probably how the mixer intended it. Herein lies the same issue: more and more mixes are done by the songwriter these days, and there’s typically room for improvement. This isn’t to say that the stereo field is bad and needs to be “fixed”, but rather that it can be better in a way that makes the overall song better. I’m speaking from the standpoint of someone who masters other people’s music here, so the concept may not fully translate, but I will do my best.
For stereo field enhancement, the main thing I like to do is tune the mid/side bands. I currently use the Synapse GQ-7 Graphic Equalizer for this step. It’s an incredible tool that makes it easy to process your mid and side bands separately. On every mastering session, without failure, I have two of these in a row on my pre-master bus. The first one is set to only affect my side band, and the second one only affects my mid band. With these two equalizers I like to inversely EQ the mid and side bands, creating some width in the track. I commonly find that subtly boosting the high end of the side band and proportionally cutting the mid band in the same area creates some nice “air” in the track, brightening it up a bit without becoming harsh. Thereafter I commonly will do the opposite: boost the low end of the mid band and cut the low end of the side band. These actions together brighten up the high end of the spectrum and solidify the bassiness of the song.
Bonus tip: cutting the low end out of the side band ensures that the bottom of the track is mono, and the differing EQ curves on the mid band and side band widen out the track and make it seem more spacious.
I would recommend shying away from stereo imaging plugins during the mastering process. You can do it yourself manually with better control, and it’s easy to make a track sound unnatural with these types of tools.
Bonus tip: I use a mid/side polarity inversion trick on every one of my masters. The pre-master bus comes in handy here, because I make two duplicates of it, for a total of three mixer channels. One of them plays the source file with the EQ and stereo field enhancement, and the other two are hard-panned left and right with muted mid bands. What’s left is the usually ghostly-sounding side band on each duplicate, which I then cut the low end out of. Finally, I will invert the polarity of one of these duplicates and mix them in with the source channel quietly. What you get is a huge reinforcement to your stereo field - the inverted polarity of one of the duplicated side-band channels creates a nice 3D effect that collapses to -inf in mono. When mixed quietly, it sits behind everything else and is a nice psychoacoustic effect that makes your master seem extremely spacious, but still clean.
We’re breaking all the conventions in this article. By “sheen”, I mean subtle reverb. Not every song will need this, but almost every song can benefit from it if done properly: it can be exceptionally tasteful and really make a good mix shine. Many engineers will scoff at the idea of adding reverb during a mastering session. I don’t blame them: it’s extremely easy to ruin a good mix with master reverb. However, we aren’t putting it on the master channel, and we will be carefully taming and tuning it.
The first thing to keep in mind is that this reverb will be added to a parallel channel. This means I commonly have yet another duplicate, bringing the grand total up to 4 mixer channels before the master channel: one source channel, two for the mid/side polarity inversion trick, and this final one for reverb. It’s exceedingly important to ensure that your reverb remains subtle and doesn’t have any early reflections or pre-delay in this stage. Early reflections can clash with transients and pre-delay can make the master sound too washed out. Furthermore, I like to keep a small decay and high diffusion on any master reverb I’m applying. This reverb should be mixed in quietly with no low end, and sit way behind the mix, resembling the volume of the two side band duplicates from the previous stage. If done right, this reverb channel can be thought of as something like shading on a detailed sketch. The song is obviously the drawing, and when we add in this “shading”, it should only serve to fill in the back of the mix a bit and give it some further depth and detail.
Sometimes I will get a very flat mix to work with. Telling my client to go back into the mix and fix it would be laborious and we would end up spinning wheels and going back and forth for far too long. In cases like this, the bulk of getting a good master relies on the “enhancing what’s there” stage, and reverb plays a huge part in this. In extreme cases, I will employ “staged reverb” to get the sound my client and myself are looking for. This entails multiple duplicates of master reverb with differing settings, all subtle and quiet, and each seeking to doctor a certain aspect of the mix. For example, if the vocals are flat, a reverb channel high- and low-passed in on the vocals with a small plate reverb can add some depth to them. Say this same mix also has an issue in the high end: the effects channels are boring and lack depth. Another duplicate of this reverb channel (high-passed all the way up to the problem area) with the same reverb algorithm (to retain cohesion), but slightly different settings, can fill out the high end in a subtle way and tie the added reverb channels together. Obviously this isn’t a perfect solution, but short decay times and quiet volumes will prevent the mix from sounding like you’ve slathered a ton of reverb on there and called it a day.
Chapter 3: Reducing the clutter
Every action taken during chapter 2 should have been extremely subtle and careful. Each action alone is likely not distinguishable, but the sum of all of your small changes should make the mix sound “better” in a side-by-side comparison. More depth, better width, a more musical EQ curve - all good things and done in good taste. This chapter, however, is about cleaning things up a bit and preparing the file for your final master channel processing. All of the coming changes take place after any chapter 2 processing: everything forthcoming is separate and does not interfere with any processing that’s already taken place. To achieve this isolation, I sometimes bus any duplicate channels made during chapter 2 together into some sort of “merge” channel, and run that into a separate channel for the chapter 3 processing. It becomes a new pre-master, if you will.
Removing resonant frequencies
The first thing I listen for when moving into this stage is the presence of “resonant frequencies”. All of the processing done in chapter 2, while subtle, can introduce a buildup of frequencies that introduce muddiness. Usually, if you solo the side band at this point, you can hear what I’m talking about - overlapping reverbs or reverb interfering with your stereo field enhancements, and so on. It isn’t only relegated to the side band, but this is the area where you’ll have the highest probability of running into the issue. The simple solution is equalization with a high Q value. Don’t be shy to cut by a few dB or more if needed, and use a spectrum analyzer and the high Q value to zone in on the problem frequencies.
It is important, however, to understand that not every resonant frequency you hear will be real (as in not present in your audio). You can get false resonants in the low end, for example, due to your listening environment. Your listening environment will, depending on its size, have a collection of resonant frequencies termed the “room mode”. Your room will have nodes and antinodes, which cancel and amplify certain frequencies respectively. The easiest way to explain it is this: any sound whose wavelength matches the dimensions of your room will be resonated by the room. This happens all the way up to a certain frequency based on your room’s dimensions called the “Schroeder Frequency”. For example, in a room 16x20x10 feet, your Schroeder Frequency will be at 163Hz. Any sound up to this frequency with a wavelength that matches this example room’s dimensions will be resonated. This antinodal behavior can cause you to cut too much out of a master to compensate, resulting in a master that doesn’t translate very well in other places.
During this stage, I will use multiband compression to take control of sibilance and plosives. A “sibilant” sound is typically a sharp consonant sound, most commonly associated with “hissing” on “s” sounds in vocals. However, you can get sibilance on any mix element, whether it’s your drumset or a sharp guitar sound. This isn’t exactly in line with the definition of sibilance, but in the audio world, it’s sometimes used interchangeably with these harsh tones. “Plosives” are similar to sibilant sounds, but these are typically that short burst of air you get on “p” sounds or “t” sounds.
Most de-essers can handle both plosives and sibilance, but I find that these tools are commonly a bit too strong to use on an entire master. Again, this is ideally something that should be fixed in the mix, but sometimes that luxury is not afforded to us. So what do you do? I find that careful multiband compression can take care of these issues with ease. Zone in on the problem areas and lightly compress. It isn’t the best solution, but as mastering engineers, we have to work with what is given to us.
This step is short and sweet. On every mix I’m given to master, I will low-cut between 20-35Hz and high-cut between 19-20kHz. Low frequencies have the most power, so you can inch a little more headroom this way, and most people can’t hear the frequencies you’re high-cutting. The low-cut gives you a bit more room to work with later on, and the high-cut can help reduce harsh frequencies and prevent audible distortion down the line.
Cleaning up the spectrum
Finally, I often find myself employing some sharp, drastic EQ cuts during this final stage of clutter reduction. Very high Q values (all the way up!) and carefully selected frequency ranges that need the most work. A good mix may not need anything here, but it’s still a good idea to sweep around and carefully listen after giving your ears a break. You may hear some harsh tones or unwanted frequencies that we can do without. This section too is short and sweet, but important nonetheless.
Chapter 4: Master bus processing
You’d think that an article detailing the mastering process would have actually talked about master bus processing by now. Everything up to this point has been advice on mastering in general, and can be tweaked to work with any track. However, to delineate the actual master bus processing side of things, I’d like to be a bit more specific. The following is a snippet from a “before” version of one of my clients’ songs:
Having already mastered it, as you’ll see at the end of the article, this song (but not this particular clip) has been through everything from chapters 1-3, and to proceed, I’d like to lay out the particular master chain I used to get it where it is now.
The very first thing I did after “enhancing what’s there” and “reducing the clutter” from this mix was run it through three separate limiters on the master channel. Yes, three. I understand that it sounds excessive, but that’s because people associate limiting with hard gain reduction and driving signals to get a lot of volume. Limiters by their very nature are tough tools - compressors with infinite ratios that can smash your mix into submission. However, I find a bit of finesse can go a long way.
The trick is to allow each limiter to work on only a section of the signal, and to not work any one limiter too hard. For this particular master, I had my first limiter set with a fast attack and fast release. This limiter contributed less than 1dB of gain reduction, and the gain reduction meter was “dancing” with the groove of the track. The second limiter had a similar gain reduction, but medium attack and fast release. The last limiter, you guessed it, slow attack and fast release. Each limiter was only catching peaks and working on its respective section of the signal due to the differing attack times, and each limiter was only limiting a dB or less. This careful taming of the signal through each respective limiter allows for a very controlled grip on the dynamics of the track and as you’ll hear later, allows the track to reach a commercial volume.
After my series of limiters, I ran the signal into a soft-clipper. A soft-clipper works by “rounding off” the peaks of a signal past a user-defined threshold instead of “squaring” them off as in hard-clipping. It is a softer form of clipping that usually doesn’t draw too much attention, and can be a great way to further tame your peaks and squeeze some volume out of a track.
Again, as with the previous stage, the soft-clipper here was barely touching the signal. Gently taming the peaks of a track over time is the name of the game. I find a good way to explain what’s going on here is by likening this stage of my master chain to “sanding” down the signal a bit. This gentle control of peaks over time will allow for a very clean and clear master later on.
I know I said earlier in the article to shy away from stereo imaging plugins during the mastering process. However, this stereo imager in my master chain isn’t there to widen anything out. In fact, it’s here for the opposite reason. I set a crossover point down around 150-200Hz and collapsed it to mono. This was taken care of earlier with my Synapse EQ’s, but it doesn’t hurt to do it again in case any of the processing thereafter introduced stereo elements down there.
While we’re on the topic, it doesn’t really hurt to have stereo elements in the low end if your track is mixed properly and you know what to listen for. People say to always make sure your low end is mono, and it is good advice. However, there are a ton of examples of commercial songs with stereo basses, containing plenty of side band info in the low end. The issue to look out for is phasing - if you have stereo bass information, when collapsed to mono, it may conflict with itself and cancel out. It likely won’t cancel out completely, but it can detract from your low end energy and cause translation issues. Because of this, I find it worthwhile to simply make the low end mono while I’m mastering to ensure there are no issues.
For this particular track, I sent the audio through an 1176 compressor and then an LA-2A compressor. I used software emulations for this track, allowing more freedom when it comes to the original settings than the hardware counterparts had to offer. The 1176 was set to a 1.1 ratio and had a constant gain reduction, dancing with the track, of around 2-3dB. This compressor adds a very bright timbre to the song that I felt really accented it and tied a lot of my prior processing together. The LA-2A thereafter had a flickering gain reduction of around 1-2dB, and the only way to describe what it does is that it “fattens” up the sound. A nice warm character is imparted by this unit, perfectly complementing and evening out the tone added by the 1176.
The purpose of this stage in my chain was primarily the harmonic info and EQ characteristics added by these units. However, they do serve a larger purpose: the 1176 helps to even out larger peaks and troughs in the signal, creating a more consistent master, and the LA-2A brings down the peaks that the 1176 brought up by extension. That may seem counterintuitive, but the way it’s working in this scenario is that the 1176 brings the entire signal up, and the LA-2A evens out unintended consequences of this. I believe that these two compressors balance one another out in both timbre and dynamics, and make a perfect team.
Multi-stage limiting again
Yep. You read it right, more limiters. This particular artist wanted a loud and clear master. Something that rivals other songs in a DJ set. To deliver, I had to run it through another set of limiters. They’re essentially the same as the first three, and each only limit by a dB or less. On my first limiter, I had a fast attack and slow release, followed by limiter 2 with a medium attack and medium release, and capped with limiter 3 at slow attack and fast release. You’ll notice that these attack and release times are all different from one another and are different from the first set of limiters. After the compression and soft-clipping, I found it tasteful to carefully apply these limiters again, working on different aspects of the signal. This gave me a couple more dB to work with and allowed me to reach my client’s desired commercial volume.
I reiterate, careful taming and sanding down the signal through multiple units all helping in their own small way is what makes a great master. Of course, this is my opinion, and there are a ton of different ways to approach the mastering process. However, for this track, I felt this was the best method so far.
At this point, the song was almost where it needed to be. However, I felt that there were some areas of the spectrum that could be tweaked a bit after all of this compression and limiting. Even though each limiter and compressor thus far has only been affecting the signal a small amount, the subtle changes add up and this can lead to mud in the spectrum. Here I had two separate stereo EQ’s: another instance of Synapse GQ-7 and an emulation of the Trident A-Range EQ.
The Synapse EQ was used to attenuate areas of the spectrum, allowing a bit more headroom before my final processing. Small cuts here and there with narrow Q values allowed me to reduce some of the background elements that were brought up in volume due to the compression and limiting. Thereafter, I found a spot in the spectrum with a boomy bass note that wasn’t attributed to any room antinodes. Reducing the EQ at that value by a couple dB cleared up the track a lot, as the bass was overpowering the low-mids and mids a tad.
The Trident A-Range EQ was in the chain only for the built-in saturator. Allowing the saturation light to flicker slightly with the groove of the track sharpened the sound up in a small but noticeable way. These added harmonics are extremely subtle, but serve to accent the track and add a small amount of further clarity to the song.
Adding a soft-clipper to the chain again at this point is more of a safeguard. It’s there but it doesn’t really do much. Only at the loudest and most climactic parts of the song will it actually clip anything, and even then it was only working by sub-dB amounts. I believe the most soft-clipping gain reduction on this unit was .5dB.
One may ask what the point is if it’s doing so little. Truth be told, the chain likely would have been fine without it. This was the last unit before my final master limiting and if nothing else, it allowed the final limiter to work a little less. I’d rather have a soft-clipper take on a bit more work than a limiter, as they’re a bit more delicate and musical.
The track at this point was where it needed to be. No more gain increases were necessary and the spectrum was clean and clear. This limiter at the end is only there to catch the occasional peak that pops out and provide intersample limiting. The easiest way to describe what an intersample peak is is to say that sometimes your volume meters aren’t quick enough to tell you your actual volume. If you’re working at a standard 44,100 sample rate, this means that your signal is broken into 44,100 samples per second, and these sample points will not always land at the top of your waveform. A limiter with an intersample peak option will use adjacent samples to find the true peak of your signal and limit based off of that value.
This was the final stage of the master channel processing for this particular track, and here you can hear the “after” comparison (with “before” for accessibility):
Chapter 5: Automation of master bus elements
Sometimes it is desirable or necessary to automate your signal during the master process. I typically find myself only automating either volume or EQ, however. Volume automation can be used to create tension and release or to fix audible distortion arising from your master chain.
For example, more “round” sections of a track may not translate very well after soft-clipping. By “round”, I mean areas of the signal with heavy sine waves or square waves. I know a square wave isn’t “round” by any means, but the flat top of the cycle is very sensitive to soft-clipping, just like the already-rounded top of a sine wave cycle. In cases like this, volume automation may be necessary - lowering the volume before your master channel processing will cause each unit to work less, and in turn, your soft-clippers to work less. Overall, this typically fixes the issue and I find myself utilizing it in almost every master. On the other hand, it may be desirable to increase volume before the master chain in same cases. Intros and breakdowns that are much quieter than the verse/chorus/drop can be brought up for consistency or simply for taste.
Bonus tip: I find it incredibly powerful to automate the volume and stereo width of the signal down slightly as a rise or transition is occurring. When the chorus/drop hits thereafter, I automate the volume and width immediately back to default. This juxtaposition of volume and width creates a huge sense of release, and as most will say, a good track is all about tension and release.
The sudden increase in volume and width creates a more impactful chorus/drop
EQ automation can be used in very much the same way as it is used in my guide on mixing: a static EQ change may not work for the whole track, so the change can be made only during the appropriate sections. For example, a low-end boost to the whole signal during the intro and breakdown can add a lot of power and warmth to the track. However, if that low-end boost were static and was kept during the chorus/drop/verse, it may sound muddy and overpowering.
Another common example that I find myself executing is either attenuating or boosting the signal when vocals come in. It may be desirable to boost or cut certain frequencies when they’re present, but that boost or cut doesn’t translate well when they stop. This can be used to accent the vocals and bring them to the front, or to eliminate harshness or sharp tones resulting from the channels.
Finalizing the master
Reaching this point can feel like a breath of fresh air. Sometimes you just can’t get a mix to do what you need it to do during the mastering process, resulting in headache and stress. However, it’s very rewarding to get the master where it needs to be and say that you’re ready to export.
Listen all the way through
It’s tempting to jump right to the export button after you’ve had a section of your master on loop trying to get the settings right. However, a common mistake people make is just that: they only master a loop and then export the whole song. Yes, the master channel processing will still apply to the whole song even if you’ve only had a segment of the track on repeat while tweaking, but there may be issues that arise due to your settings in other places. Listen for audible distortion/clipping, listen at different volumes, listen against a reference track, and most importantly, listen after you’ve given your ears a break. If it still sounds good in an hour, set your export settings.
Bonus tip: Pay particular attention to sections that come after silences in your song. Silence resets each unit in the master chain, and if they all activate at once you can get unwanted artifacts.
Set export settings
This is very much the same as setting your export settings during the mixing process, except you’re going to dither this time. Dithering masks quantization noise, so it’s important to remember. Your export settings will heavily rely on the delivery medium of your song. If it’s for personal use, a standard 16-bit wav, aif, or even 320kbs mp3 can be fine. If you’re sending your music to a reproduction house to be burnt to CD’s or cut to vinyl, you will want to ask them what they prefer. A client of mine sends an album a year that gets cut to vinyl - their reproduction house always asks for 24-bit non-dithered audio, for example. Once this is all set, hit export!
After your mastered file is exported, you may want to embed metadata. This info is generally your artist name, year of production, album name, etc. There are several programs (free) that can do this, and I highly recommend it. Furthermore, if your music will be played on the radio or distributed via any sort of streaming service, you may want to purchase and embed ISRC - International Standard Recording Codes. To simplify it, these can be purchased in large chunks and both help to identify your music and ensure more accurate royalty payments.
As with my mixing article, I understand that this is a huge chunk of information to take in. After everything said here, it’s very important to keep in mind that every master will be different and you should approach it as such. My best advice is that if anyone tells you how to master a track, question everything about their input. You can't throw it through a mastering preset and call it mastered. You can't "use [this] compressor with [these] settings", or "always do [this thing] with your limiter." These are hackneyed tidbits of info that get thrown around because it makes a track a bit louder and brighter. This is the reason it’s difficult to make any surefire guide on the topic. All I can do is walk you through my process, as I’ve done here, and hope you take something useful from it.