Hello [dot]cool glitchers,

This is the second post in a series of 3 posts I'll be making that I hope will address a few common concerns and questions I see producers ask regarding mixing, mastering, and getting your tracks to sound consistent across multiple playback mediums. The focus of this article will be mastering your music to achieve competitive loudness, modern clarity, and great audio translatability.

I will be discussing some techniques and discoveries I’ve come across when mastering my latest track: 'Fractured Silk' (releasing soon!). It may be helpful in your own mastering endeavors, especially if your music falls under more experimental and generally loud electronic music.

A few notes regarding how I decided to present this; It’s not super feasible to get exactly what each bounce sounds like, at full volume, through my DAW as Ableton will generally let you clip a bit the master a bit before noticeably distorting the sound. As such, I decided that the best way to present exactly the effect each plugin had on the sound of the track, was to bounce each out normalized except for the final limiter. The loudness may differ between bounces but I’d recommend to take a listen to what changes each plugin made to the sound overall instead of the perceivable "loudness" within each clip.

Final Note: For the most part, these effects were applied non-linearly which means that I almost ALWAYS had a limiter or a compressor at the end of the chain while adding and tweaking parameters. This meant that during each stage of plugin, I had an intended dB target of limiting in mind that I wanted to hit (2-4 dB of gain reduction on master limiter), and would adjust each plugin to taste with that constraint in mind.

Please follow the link at the very bottom of the page for the isolated bounces for each step of the chain. They are numbered in sequential order which is the order they will be discussed for each section:

That being said, here’s a list of what’s on the master.

What Is On The Master Chain

1. Clean Mixdown (No effects)

2. Oxford TransMod

1a. -5.15 dB input gain since mixdown is too loud lol
2b. Ratio +0.044
3c. Overshoot 5.727ms
4d. Overdrive 1.14%
5e. Rise 0.43ms
6f. Recovery 70 ms (Ratio from bpm @ 140bpm)

3. Multiband Dynamics

1a. Frequency/Input dB/Output dB
2    i.     High: 2.59 kHz 	/ 0.00dB  / -0.57 dB
3    ii.    Mid: noFreqValue	/ 0.00dB  / -1.42 dB
4    iii.   Low: 119 Hz		/ 0.00 dB / -1.91 dB
5b. ‘T’ (Time parameters) Att/Re  l
6    i.    35 ms / 1400 ms
7    ii.   70 ms / 143 ms
8    iii.  72 ms / 143 ms
9c. ‘B’ (Below thresholds and ratios) dB / Ratio
10    i.    -32.2 dB / 1:1.00
11    ii.   -48.1 dB / 1:1.09
12    iii.  -40.0 dB / 1:2.00
13d. ‘A’ (Above thresholds and ratios) dB / Ratio
14    i.    -23.1 dB / 1:1.01
15    ii.   -22.9 dB / 1:1.18
16    iii.  -20.9 dB / 1:1.43
17e. Output: 13.79dB  
18f.  Time 100%
19g. Amount 100%
20h. RMS mode ON
21i.  Soft Knee: OFF

4. Hammer DSP (and Pro-Q3)

1a. LO Cut enabled (Basically just a 6 dB/octave roll off at 80hz as indicated [by the manual for the VST](https://objects-us-east-1.dream.io/kushfiles/manuals/USER%20MANUAL%20-%20Hammer%20DSP.pdf))
2b. ~5.3 dB boost @ 30hz
3c. 1.8 dB boost @ 1.6khz
4d. ~2 dB boost @ 15khz
5e. Reversing polarity of entire signal switch: ON

5. Pro-Q3 Settings:

1a.   +1.43 dB Lowshelf @ 61.768 Hz, Q set to 4.430.
2b.   -0.6 dB Bell shape @ 82.328 Hz, Q set to 3.900
3c.   +1.21 dB Bell shape @ 109.98 Hz, Q set to 3.690
4d.   -0.33 dB Bell shape @ 146.90 Hz, Q set to 13.03
5e.   +0.36 dB Bell shape @ 247.10 Hz, Q set to 4.330
6f.   +0.21 dB Bell shape @ 329.63 Hz, Q set to 2.666
7g.   -0.26 dB Bell shape @ 412.48 Hz, Q set to 3.664
8h.   -1.93 dB Bell shape @ 522.15 Hz, Q set to 6.130
9i.    -0.34 dB Bell shape @ 830.01 Hz, Q set to 14.50
10j.    +0.14 dB Bell shape @ 881.90 Hz, Q set to 5.928
11k.   +0.21 dB Bell shape @ 1315.6 Hz, Q set to 5.043
12l.    -0.09 dB Bell shape @ 2091.7 Hz, Q set to 4.690
13m.  -0.21 dB bell shape @ 3175.0 Hz, Q set to 2.898
14n.   -0.26 High Shelf @ 3951.1 Hz, Q set to 2.900
15o.   -0.39 dB High Shelf @ 10481 Hz, Q set to 8.690
16p.    Low Pass @ 22000 Hz, Q set to 1.000

6. MSpectralDynamics UI screenshot.

Nothing too fancy going on here I promise; The UI is daunting but really the only thing that’s occurring is one instance of compression @ -22.9dB with a 1.18 ratio with dynamics detection and spectrum settings to taste. I ended up turning on the 2nd processing at -20 with 1.43 downward compression ratio as well for just a bit of extra processing roll-off throughout the compression shape and for less obvious linear rolloff.

The Threshold holds a relation against the set multiband compression. I find that if you want to utilize a lot of slight compression in a long chain, especially when utilizing multiband compression, it’s very useful to compare to and assign similar ratio/threshold settings so that you don’t have “stacks” of compression but rather layers. “Stacky” compression sounds rather obviously compressed in tiers and you have these very obvious differences in dynamics when the signal hits certain compression thresholds and sweet spots. By noting the previously utilized compression ratio/threshold, you can effectively double up on the compression while still maintaining a sort of relation between the two compressors which I find tends to sound a lot louder but also more natural and “glued” instead of heavily contrasted between dynamic and squashed.

Note the ‘naturality’ knob setting was in place to not create too much of a spectrally warble in the low end. Default settings are 5%; I ended up using around 2.1 since I found the FFT bands utilized hits a nice sweet spot between “wide enough to not be tonally resonant” and “small enough to be precise”

Input/Temp gain parameters set to taste. This can be done with any other compressor by adding a utility (+ X gain) before the compressor and adding a utility (- X gain) after the compressor. It essentially acts as a knob to control the amount of compression applied.

Smoothness at 0.43% to give that extra bit of silky-ness to the compression and blend the FFT bands utilized so the gain reduction is a bit smoother feeling. I found this to be a very sensitive parameter and a little goes a long way.

66% Dry/Wet applied as I found that 100% wet ducked the high end in a nasty way and 0-50% wasn’t attenuating the sub enough.

At its peak, this compressor is reducing dynamics by no more than -1.8dB (during a particularly nasty point of resonance) but for the most part is hovering around -.25 dB. Mostly unnoticeable but is hugely beneficial (in my opinion) when preparing for the master limit.

7. Oxford Inflator

1  Note: (Can be VERY easily replicated using MSaturator; It’s just a really easy to use waveshaper)
2a. Input: 2.69 dB
3b. Effect: 9.6%
4c. Curve: -20.8
5d. Output: -2.69 dB
6e. Clip at 0 is OFF
7f. Band Split is ON
8    i. Important to note this because I believe it’s applying separate waveshaping for each band which I found to be far nicer sounding than a flat waveshaper over the whole master.

8. Standard Clip

1a. Clip gain: -1.63dB
2b. Hard Clip
3c. No oversampling
4d. Ceiling: Off
5    i. A note on clipping: This was mostly used to cut down on barely noticeable transient spikes. Helps to reduce the load on the limiter input and sounds a bit smoother with it on. I did not clip more than -0.3 dB or so on “intentionally loud” moments. 
6       That is to say, the kick/impact moments have a touch of clipping to keep up punch and presence while also clipping some of the more clicky/crackly transient moments on the track.

9. Oxford Limiter

1a. Input Gain: 7.216 dB
2b. Attack: 0.0542 ms
3c. Release: 0.089 ms
4d. Soft-Knee: 0 dB
5e. Safe Mode OFF
6f. ‘Enhance’ = 0
7g. Output Trim: -0.523 dB
8    i. Mostly in place to not clip the limiter; Intersample peaks are very present throughout this master lol. Generally I like to bounce out the master with -.2 dB of headroom to account for streaming service compression.
9    ii. Additionally to ensure that any peaks that may have bypassed the limit (by being a small enough attack time, less than 0.0542 ms).
10h. Auto Comp: OFF

What Does This Mean?

With all of the plugins/parameters listed, this definitely seems like a lot but I promise you that what is actually occurring is a bit simpler than it may appear.

1. Transient Enhancement (TransMod) Used to increase the punch and power of impact moments as well as ensuring the snare punches through the mix nicely. Slightly bricked but diverse frequency density glitch moments also have a bit more depth given to them.

2. Multiband Dynamics Attenuation Reduces the Bass/Sub Balance while compressing any large peaks in the mid range. Also introduces very slight compression on the high end. Slight as too much will really kill the “air” in the master.

3. 3 Band EQ and Pro-Q 3 Tightens up the sub bass while adding a bit of “presence” at 1.6khz which helps with translating the mix across multiple systems. It also provides the high end with a bit of shimmer. The Pro-Q 3 also helps to finish off the mix with a bit more sub while also attenuating the immediate area above it; Helps keep the sub frequencies less “boxy” sounding.

4. Compression Targeting low end peaks and tightening up the low end to reduce “Boom” and “flab” going into limiter Slight multiband waveshaping/Distortion Very slight curve/waveshaping; with less than 1/6th of the signal being effected.

5. Final Limiting

I’d also like to stress that while I tend to use a lot of 3rd party plugins, I believe that many/all of these can be replaced with stock plugins or free plugins as well. My plugin selection is purely due to familiarity and ease of use more than anything. From my collection of 3rd party plugins, similar effects can be achieved by using stock plugins:

TransMod: Can be replaced with Ableton's drum bus with no distortion and no decay on the ‘boom’ parameter or a free 3rd party plugin like kilohearts’ transient shaper.

Hammer DSP can be replaced with an EQ3 and an EQ8 (which could handle the roll offs). Any other native EQ plugins would also work just fine.

MSpectralDynamics may be the exception here; I’m fairly sure multiband dynamics or glue compressor or compressor could work here. All it’s really doing is attenuating peaks in bass/sub frequencies which can theoretically be done with a regular compressor. You may need to verify via spectrum or SPAN or something that the bass/sub is hitting the compressor first though. Glue compression would most likely provide the closest effect for what I’m using MSpectralDynamics for though. Keep in mind it’s very slight with only .25dB of gain reduction overall.

Sonnox Inflator can very easily be swapped for MSaturator (free). Here’s another fantastic video explaining exactly what Inflator does. My use case is very similar to the behavior demonstrated with ‘clip at 0’ turned off.

The Deep Dive

As noted in the intro, I generally had a limiter on the master channel the whole time, and I would generally recommend having one to start with. You can always A/B it during the mastering process but I find that it’s good to see around where your track sits, at approximately the intended gain reduction, before any additional effects.Typically, I’ll set the limiter to around 4-6 dB of gain reduction as that’s usually the target range I’ll strive for with a track that is a bit bassier like this one.

Once the limit is set, you can begin to assess the master overall. Does it feel “full” enough currently? Which areas of the mix are hitting the limiter first? Do you feel like you would want more limiting? Does the track still punch nicely? Questions like these can indicate that perhaps the master hasn’t been treated enough via other effects and you may need to apply additional compression before the limit.

I’ll usually start a master off with some kind of transient shaping; Typically in the form of Sonnox Transient Shaper as it allows me to tackle a few things at once: coloring the sound via distortion/overdrive, Transient enhancement, and messing with release times to change transient clicks into a bit more “punch”. The overdrive helps the transient enhancement sound a bit more natural and also helps round out the enhancement a bit.

Without the overdrive I found that the transient modification isn’t really heard much but notably increases the dynamic range of the tune overall; Not ideal when it comes to mastering as you’ll want to reduce the dynamic range in preparation for final limiting.

Note: I don’t think it’s explicitly necessary to do this step on the master bus; I just found that I liked the sound overall so I kept it. Transient modification is probably better off to be done on a channel/bus basis so you have some more nuanced control over transients (such as different transient enhancement for melodic elements and basses vs drum transients).

Multiband dynamics and the hammer EQ (+ just a touch across the spectrum of Pro-Q3) are doing the most preparation for mastering in my opinion. They are incredibly useful tools for slightly different reasons. The multiband can be useful, just outside of its compression ability, for band splitting and balancing a mix by mixing different bands to taste. Specifically, I found that setting the mid/high split frequency allowed a lot more of the high end to poke through (noted in elements like hi hats and snare) without bringing up the entire mid-range as well. Pro-Q 3 added at the very end of this step, is used to balance out the sub/bass frequency ratios. By adding a bit more sub, the tone of the low-end sounds a lot less resonant and flabby while still keeping focus on the energy of the basses.

With applying the compression on the mid range (via MultiBandDynamics), we can take care of any peak frequencies which may cause mud or too much dynamics (can be bad as it results in having to limit the mix more overall) in the final master while keeping the top end of the mix relatively free from compression which lets the mix breathe a bit. Compression on the low end helps keep the sub from overwhelming the mix and helps hide some of the compression on the mid range as well. From my general experience, I’ve found that compression on the low/mid tends to be a bit less noticeable than compression on the high end. As such, I recommend playing around with compression on the mid range if you’re trying to squeeze some extra loudness out of the mix; You might be surprised at how much you can get away with!

Speaking of loudness, I really love the Hammer EQ for the purpose of boosting musically. I find that the boost around 30hz with the 6db rolloff is a great way to gain some extra focus in the bass/sub. I find it’s best applied after the multiband compression as for this track in particular, ~50hz is right around where the root note of the sub for this tune is and by rolling off any subharmonics that might be present in the sub, the effect gained is more of a perceived loudness boost. With the sub sounding tighter, more focused, and more powerful. "Round" as an adjective may be applicable here as well.

The extra presence at 1.6khz and 15khz helps keep the track lively sounding and shimmer a bit. I’d view these last 2 eq adjustments as mostly “cosmetic” and give the track a bit of a nice sheen which makes it feel a bit more “finished” sounding. If your mixdown is good, these cosmetic eq decisions may not be necessary or can really push the track over the finish line. They can also help to make up for weaker areas of the mix or be used to patch any major holes that might be present in the frequency spectrum which can result in “hollow” or “bloated” masters.

The 2nd to final step is then to see how much saturation is needed to make the track feel “full” in tandem with the other effects already dialed in (including limiting!). I find that this can help you squeeze a bit more out of the limiter as well but I find that by not focusing on saturation until the end, you can get a clearer focus on how the extra harmonics generated help reduce dynamics and lift the noise floor (via harmonic generation) in a much more apparent way than if done prior to limiting and other mastering effects. It’s especially helpful when trying to dial in “just enough” without deep frying your mix and mastering it to the point where it would be better off without it. I’d say to be fairly restrained when applying saturation onto the master, you don’t wanna cook the entire thing after paying such careful attention to the previous steps. I’d say go for a final “baste” to really lock everything in while also giving a last bit of harmonic blending to the almost-finished master.

Finally, we have the last stage of loudness. These last 2 plugin choices will typically determine how loud the mix is overall. Up to this point, we’ve mostly just prepped the mix for final squashing and it’s important to not go too heavy handed at this step. An over compressed/limited mixdown can often be antithetical to the work done up to this point.

That is to say, an over-limited master can be: less punchy, quieter sounding (due to ducking occurring if any one element is taking up too much headroom), and deaden stereo width (especially true on limiters not operating individually for L/R channels). It should be avoided as much as possible. Every limiter is also not built equally. Some may be able to squeeze more than others and I think it’s worth swapping between limiters to determine which one might be best for the job.

I find that, depending on how dynamic the track is, there’s no real set range for how much you can push your final limit. I typically will push the track until I notice that the limiter isn’t backing off entirely after transients or impact hits (basically overpush the limiter to find the point of over-limiting) and then dial it back around 1-3 dB. This can get you into a nice ballpark range for limiting and here’s where I would say is a good point to assess the previous steps taken.

Another way to dial in a limiter could be to have a set number of gain reduction in mind (say for example: 6 dB) and then boost as much as possible without going over that threshold. Then you can make slight adjustments (increments in either 0.1 or 0.3 dB I find tend to be fairly safe). You should notice that at this point, the adjustments should be fairly audible within the mid range. I’ll typically stop adjusting the input gain on the limiter when I find that the peaks are stably hitting the intended gain reduction target and that the loudest parts of the track don’t push the limit over that threshold. I find that reducing by another 0.1 dB or so after hitting the “perfect” limit is also decent practice to compensate for any ear fatigue that may be present as well. As always, make SURE you take another listen after giving your ears a break as well. :)

Closing Thoughts and Clip-To-Zero

Clip-to-zero is a really interesting trend that seems to be sweeping the experimental bass scene. I think the results speak for themselves as you can get some absolutely BONKERS amounts of loudness out of tracks by (in some cases) completely omitting a limiter and just clipping instead. If I’m honest, it’s a bit conflicting. On one hand, yes insane loudness. On the other, I think that coloring a sound too much before a final master can make it really hard to judge exactly how much clipping is too much clipping, which could result in more homogeneous timbres and mask some of the finer details in the mix.

My personal philosophy tends to be that if everything else goes very well during the master stage up to the point of final limiting, and you want to eke out just a smidge more loudness, clip as little as possible then limit. I’m hesitant to throw out a number but I generally prefer <1 dB of clipping.

In this particular track, clipping was applied just before limiting to take care of a few odd spikes in volume that came up when applying bus effects to the bass group. I consider it borderline negligible and can very barely tell the difference with it on vs off. I didn’t want people to hone in on the use of clipping as the “final touch” and the reason this track sounds loud; It’s really not. It’s inclusion is purely a debug for fx used during mixing, arrangement, and sound design.

This was a fairly technical dive into the process I'd go through when mastering a track, however, I'd like to stress that a lot of these decisions were made mostly naturally. I did not have a blanket set of tools I intended to use right from the get-go, however, this track in particular made use of very nearly every tool I'd typically utilize when working on a master, so I felt it would be good to sort of explain my intentions a bit. To take a break from the more technical aspects of mastering, I'd like to close this out with a few ideas that I believe are key to good masters:

• Create a Master that emphasizes the best parts of the mixdown and do not master to fix a mixdown.
• Be a bit careful when applying things like EQ or distortions across an entire mixdown to master; You can possibly color the mix and create an overly homogeneous array of timbres by adding too much saturation or warmth
• Clip only when necessary and do not push the clipper too hard. Be careful to monitor which frequency range is taking up the headroom in the clipper as well; Clipped subs should be avoided.
• Use multiband solo channels to monitor exactly what/how frequencies are getting affected by compression, eq, saturation, and final limiting (essentially the mastering chain as a whole).
• Pay attention to the relational loudness between the sub and high end frequencies
• Compressing the mid range can yield very dramatic increases in loudness and can increase how hard you can push a limiter overall. It's also good to keep in mind that lower to mid frequencies take up more headroom when compressing or limiting!
• Be careful not to over limit the mix and be sure to reduce the output of the limiter, post-limit, to accommodate any peaks that get through the limit as well as ensure that the final master isn't clipped after export.
• When using a reference mix, pay attention to the valleys and which frequency ranges they occupy. Peak values basically just mean that those points in the frequency spectrum are hitting the limiter first. Valleys can give a much better understanding and impression on mix down and compression levels across the frequency spectrum.

Thanks for Reading! If you have any questions, comments, or just want to learn more about audio, I highly recommend checking out our discord for glitch.cool. Our #audio-production channel has tons of great insights and goodies from myself and the glitch.cool community at large. Hope to see you there! :)