In this video, the chapters are in no particular order, but we will be discussing various forms of compression in-depth, as well as introducing some useful techniques you can use when mastering.
We’ll cover how to glue a mix together with compression, as well as how to very transparently compress, to get a cohesive and colorful sound or solely control dynamics respectively.
Then we’ll cover how compression can alter the frequency response, and how we can use an MB compressor to mimic how our ears compress in order to create natural-sounding compression.
Next, we’ll discuss New York-style parallel compression and how it can be used creatively to affect a master in a way most regular processors can’t before we discuss the benefits of simultaneous expansion and compression, and how compressors can be used to alter a master’s stereo width.
Next, we’ll cover how to bring quieter details of a master to the forefront without the need for aggressive peak-down compression.
Lastly, we’ll cover how to de-ess during a mastering session to attenuate sibilance, but avoid compressing unrelated frequencies, how to reduce resonances with a couple of different processors, and finally how to use dynamic EQ to attenuate out of key frequencies.
I’m guessing you won’t want to use all of these types of compression, but at least 1 will definitely be helpful for your style of mastering, so stick around to learn about each type and technique.
One of the most popular ways to introduce compression when mastering is as a way to glue together the sound. Glue compression is created by setting a longer release time so that multiple transients are compressed collectively.
The benefit is that the master sounds cohesive, but we need to be careful since affecting multiple transients can affect the perceived timing of the track.
To do this the right way, we need to time our compression with the BPM of the track. So if our BPM is 140, we can divide 60000 by 140 - the result is 1 quarter note in milliseconds; In this example, 429 milliseconds is 1 quarter note.
Then I can multiply this number by 4 to get 1 whole note, then subtractive the value of our attack time, and use this value as the release time. For the sake of understanding this better, let’s say we use a release time that is not in time with our track, say 1 second. Then whenever the signal returns to unity, it’ll sound amplified relative to what was just attenuated.
In other words, it’ll serve a somewhat similar purpose to a transient - if this return to unity is out of time with the track, we can severely mess up the perceived dynamics and subsequent timing of the master.
In addition to timing our release time, we can use a compressor that imparts a specific tone to the signal - this means a compressor that introduces harmonics and maybe some subtle equalization. For this reason, many engineers like to use variable compressors, more commonly called Vari-mu compressors.
I’ll use this Pulsar Vari-mu plugin to demonstrate the idea, but Klanghelm offers a great free option if you’re interested.
Let’s take a listen to glue compression being introduced with a Vari-mu type compressor, and with the release set to be in time with the track.
Sometimes when introducing compression, you don’t want to create a colorful sound - or even be able to tell that the compressor is there at all.
When this is the case, we’ll want to use clean, transparent compressors, and with in-time settings that return the signal to unity quickly, but set in a way that doesn’t result in distortion.
Let’s use this free plugin - Tokyo Dawn Lab’s Kotelnikov, to introduce transparent compression.
Notice that with it, we can introduce both peak down and RMS detection - RMS or root-mean-square detection will average the incoming signal to measure an average loudness.
It’ll result in more transparent compression; however, it may let loud peaks through, so we may want to find a good balance between peak and RMS detection, and pay attention to when the signal is being attenuated.
We’ll want to set an attack time that’s no quicker than 10ms, and a release no quicker than 50ms - the reason being, if we set quicker times, we can cut into a transient, in turn causing harmonic distortion.
This is great to have the transients stick out, but it’s not useful for clean-sounding compression.
Additionally, I’ll set the release to be in time with the track, similar to last chapter, but instead of a whole note, I’ll use a 1/16th note. To find that take 60000 and divide it by the BPM, and then divide by 4.
Lastly, we’ll want to set the stereo sensitivity toward the center or mid-image - this way we’re detecting and affecting where the majority of the master’s energy is.
Be sure to carefully set your threshold and use a low ratio to ensure we don’t compress more than 1.5dB.
Let’s listen to this compressor being used, and notice how it cleanly attenuates the master.
Multi-band compression is super useful if we want to control a specific bandwidth of frequencies - it can also be less noticeable than regular stereo compression for this reason.
All of the concepts we’ve discussed in the last 2 chapters still apply - like timing the release to be in time with the BPM and avoiding super quick attack and release times to avoid distortion. The main difference is that we can use fast attack and release times on the highest frequencies - since the distortion typically caused by these settings only occurs at low frequencies.
The main idea I want to introduce though is that compressing specific frequencies both controls dynamics and causes equalization.
So let’s say the track is sounding muddy and I’m having trouble hearing the vocals - I could center the compression on 250Hz. and attenuate by 1 to 2 dB. This will attenuate the frequency range like an EQ, but do so dynamically, instead of statically.
This a great way to alter the frequency response of a master, but only when a particular range becomes too loud.
Or let’s say the reverb is covering up the vocal too much, I could dynamically attenuate 2-5kHz on the side image so that the mid-image vocal cuts through whenever it’s present.
Also, I’ll use this FF-MB, but a good free alternative is TDR NOVA.
Let’s take a listen to MB compression being used in this way, and notice how it shapes the frequency response.
While we’re on the topic of frequency-specific compression, let’s cover how we can use MB compression to mimic how our ears naturally compress sounds that are too loud. I covered this in our last video on vocals but let's look at it one more time.
In short, our ears compress frequencies below 1000Hz, whenever the sound gets too loud - this attenuation is between 1-20dB, depending on how loud the sound is.
Also, the 2 structures that cause this can’t move instantly to attenuate the sound - it takes about 40ms for them to tense up or contract, and they stay contracted for about 150ms.
With that in mind, we can use an MB compressor and attenuate frequencies below 1000Hz by a couple of dB. We’ll set the attack to 40ms and the release to 150ms to mimic the response of the 2 ear structures, and if available, use a softer knee to cause gradual compression.
Be sure to avoid lookahead and any other unrelated settings.
Let’s take a listen, and like last chapter, you can use TDR NOVA as a free alternative to this plugin.
New York Style compression is a combination of parallel compression and EQ. To create it we’ll need to set up a send and corresponding auxiliary track from our channel track.
On it, we’ll insert a compressor and an EQ. We can use compression that glues the signal together or transparent settings but with a great overall amount of attenuation - since we’ll be blending this effect in with the auxiliary track’s fader, we can be more aggressive with our settings.
Then, we’ll insert our EQ - we could use a fully parametric one to really control what part of the compressed signal gets amplified, or we could use something more stylistic like a Pultec EQ to shape the sound - if available, be sure to use a linear phase setting for the EQ to avoid phase cancellation between this track and the original channel.
With the EQ, we could amplify the kick, the vocal range, and maybe some of the highs to accentuate important ranges - but what you amplify is up to you.
If you want to get even more of the sound of the compressor, we could insert an EQ before it, and drive frequencies into it. This way the compressor is having to work harder on those ranges, in turn causing compression on those frequencies to be more apparent.
This method controls dynamics, allows you to equalize the compression, and makes it easier to automate this compression by altering the aux track’s channel fader.
I’ve been using this FF-Pro C2, but let’s use this free SSL emulation by Analog Obsession and create parallel compression with a lot of character. I’ll equalize both before and after compression to emphasize important ranges and then blend in the effect.
Let’s take a listen, and notice how the master sounds fuller, and how the amplified frequencies stick out a bit more.
The Omnipressor is a unique compressor - it allows you to set levels for compression and expansion simultaneously, resulting in signal attenuation whenever the signal goes above the threshold, and expansion whenever the signal falls below the threshold.
As a result, the signal stays within a dynamic range, resulting in louder parts becoming quieter, and quieter parts becoming louder.
Although this effect is typically used for mixing, it can be used to great effect when mastering.
If we limit the expansion and compression ranges to 1dB, set the function or ratio near the middle, and very carefully set our threshold, we can achieve a dynamically controlled, and surprisingly loud master without unwanted artifacts.
I find that this effect works great before limiting when I want the signal louder, but I don’t want to resort to heavy amounts of attenuation to achieve it.
Although you could use 2-instances of a free compressor like MCompressor by Melda Audio to achieve compression and then expansion, the routing is different than this plugin, so unfortunately I don’t know of a free alternative that matches this plugin exactly.
Let’s take a listen to this plugin used with subtle settings, and notice how it makes the track louder, dynamically controlled and makes quieter details easier to hear.
We touched on this concept in chapter 4 when we used MB compression on our side image, but let’s go into more detail about affecting a master’s stereo width with compression.
If we use a compressor that’s capable of mid-side compression, we can introduce more compression to the mid image to cause stereo expansion - whenever the mid image is attenuated, the side image will sound louder relative to the mid or centered image.
Usually, I want to attenuate the mid-image by .5 to 1.5dB, and avoid things like make-up gain, since that would negate the stereo expansion aspect.
Some plugins that work well for this include Presswerk by Uhi, Weiss Maximizer if you select the Wide algorithm and the more comprehensive DS1 compressor, and the FF-Pro MB if you want the attenuation to be frequency specific.
A great free option though is the MCompressor by Melda Audio - if we open the toolbar section, we can change the routing from stereo to mid and side, or in this case just the mid image.
Ultimately, this is one of my favorite ways to compress a master, since it controls dynamics, and results in very natural-sounding stereo expansion.
Let’s take a listen, with more aggressive settings than usual, and notice how the master sounds wider whenever the mid-image is attenuated.
So far, all of the compression types we’ve covered cause attenuation from the peaks down, or from an averaged loudness down in the case of RMS compression.
But one of the most useful forms of compression in my opinion is upward compression. In the case of the Waves MV2 - it can measure quieter parts of the signal, control the dynamics of that quieter section, and then amplify it.
This results in a controlled dynamic range, without the need for attenuating peaks. More importantly, it amplifies quieter details and makes the overall master sound fuller and more forward.
Another option is the Sonnox Oxford Inflator - it works a little differently by prioritizing aspects of the digital signal with more 0s than 1s (at least that’s how it’s described in the manual) and then amplifying these parts of the signal, and adding harmonics to fill out the sound.
Alternatively, you could use the free MCompressor and use the custom shape setting to amplify the signal whenever it falls below the threshold - the concept is similar, but the final sound will be a little different. One more free option is OTT by Xfer records which offers frequency-specific maximization, but you’ll need to use incredibly subtle settings with this one since it’s a really aggressive effect.
Let’s take a listen to the effect with my personal favorite option, the Sonnox Limiter’s enhance function, and notice how we’re not attenuating the peaks, however, the details of the master are greatly increased, as is the overall loudness.
Now that we understand upward compression, let’s see how it can be used creatively. If we use upward compression in a form of New York-style parallel compression, we can greatly emphasize a particular range of frequencies.
To do this, I’ll set up a send from the channel track, and on the corresponding aux track insert a linear phase EQ.
With it, I’ll isolate the signal to the mid frequencies by using a high-pass and low-pass filter. Where you center these filters is up to you, but I’ll set mine to 400Hz and 4kHz respectively.
Then, I’ll insert an upward compressor or maximizer - in this case, I’ll use the Sonnox Inflator, but you can use an of the options we just discussed in the last chapter.
With the inflator, I’ll set its effect to 100% and then blend in the upward processed mid frequencies with my original signal.
With this setup, we reduced the dynamics of the mid-frequency range, greatly increased aspects that would’ve otherwise been masked, and then dialed in the effect until the overall sound is balanced.
Like chapter 6, we have the option to emphasize frequencies into the processor if we want them to be more apparent. So I’ll create a bell around my vocal’s clarity range of around 2kHz, and also amplify the side image around the 500Hz.
This is just what I thought sounded best for this particular track, but use your ear and emphasize and frequencies you want more of.
Let’s take a listen and notice how the mid frequencies are present, and how the overall master sounds fuller and more impressive.
Although de-essing isn’t always needed when mastering, it does serve an important role when it is needed.
Finding the right settings can be challenging so let’s look into it.
We’ll need to isolate the compression to the range of sibilant frequencies, usually between 5-8kHz, but you’ll need to listen to find these, or listen and observe the frequency spectrum.
After we’ve isolated the attenuation to the correct range, we’ll need to ensure that attenuation only occurs whenever a sibilant occurs. If we don’t set the threshold correctly, it’s really easy to attenuate unrelated aspects of the signal, like cymbals or other high frequencies.
Doing this correctly comes down to listening carefully, observing the attenuation, and seeing if it pairs up with when sibilance occurs, or if it’s attenuating at an unrelated part of the signal.
I’ll use this Weiss de-esser, but you could use an MB compressor. If you want a free option, T-De-Esser is a good alternative, and I would also recommend Sibilance 4 by Toneboosters, which offers a lot of control and is really affordable.
Let’s take a listen and pay attention to how attenuation only occurs whenever sibilance is present.
Although resonance reduction is usually thought of as a form of equalization, it’s really a multi-band compressor with a preset threshold that determines when to attenuate various frequencies.
It’s an incredibly useful tool when trying to create a balanced master - so let’s look at some good settings. I’ll use this Soothe 2 plugin, and to make it easy, select the Balanced to the Grammy Awards setting - which measures the mid and side separately to apply compression when needed.
From here, we can adjust the pre-emphasis EQ bands as needed, to attenuate aspects of the master that are too prevalent. The low mids are usually a trouble area - the same could be said about the sibilance range.
That said you’ll have to use your ears.
Lastly, if you use this plugin, be sure to use higher quality settings to reduce phase cancellation, and to dial in the effect with the mix dial.
Although I don’t know of a free alternative, Smooth Operator by Baby Audio is a good affordable option.
Let’s take a listen, and notice how the master sounds balanced.
In chapter 4 we discussed how MB compression can change the frequency response, and in a way, be used as equalization.
If we have a dynamic EQ, we can pinpoint which frequencies we compress - in this case, we'll attenuate out of key frequencies. This results in a more musical sound since desirable in-key frequencies become louder relative to out-of-key frequencies.
To find these frequencies, we can use this free online Tunebat analyzer, and once we know the key, look up which frequencies are in and out of key.
From there, we can use this Pro-Q 3’s piano roll to snap the band’s center frequency to exact notes, but if you don’t have this EQ, you can look up the exact frequencies of notes, and center bands on those.
Lastly, we’ll enable dynamics for each band and set the threshold as needed. We’re aiming for between .5 to 1dB of attenuation, and it’s best to set a narrow Q value so that most of the attenuation occurs on the out-of-key note.
Let’s take a listen, and notice how the track sounds more balanced, and a little more musical.