In the realm of music production, the mastery of dynamics is not just a technical necessity; it's an art.
Many of us are familiar with the concept of compression, particularly downward compression (if you aren't, then please read my article “The Complete Guide To Audio Compression & Compressors“).
However, there's another, less-discussed technique that can revolutionize the way we approach dynamics in our mixes: upward compression.
What is upward compression? Upward compression is an audio process that reduces (compresses) the dynamic range of a signal by increasing the volume of lower signal levels (“quiet parts”) while leaving higher signal levels (“loud parts”) relatively untouched, thereby enhancing subtle details of the processed audio.
In this article, we'll delve into the intricacies of upward compression, exploring its unique characteristics and how it compares to more traditional compression methods.
If you prefer video content, I have an in-depth video on this very topic, which you can watch below:
Disclaimer: This article includes affiliate links. If you click one of them, I may receive a small percentage of the sale at no extra cost to you (which I'm very grateful for, as it helps me produce more free content here at My New Microphone). You can see the list of my partners here. Thank you for your support!
A Primer On Compressor Basics
Before we dive straight into upward compression, let's quickly revisit some fundamental concepts of dynamic range compression.
Forgive me if I spend too many words describing downward compression in an article about upward compression. The best way I've found to understand upward compression is in contrast to the typical downward compression we all know and love.
As the name suggests, compression reduces or “compresses” the dynamic range of an audio signal. In most cases, this is done with “downward dynamic range compression”, which we generally refer to simply as “compression”.
Once again, the dynamic range of a signal is the difference between its “loudest” and “quietest” parts. It's important to note that, due to the attack and release times of a compressor (which I'll define momentarily), we're dealing with signal levels beyond the instantaneous amplitudes of the waveforms. Otherwise, we'd be dealing with waveshaping distortion, which is something completely different.
With downward compression, a compressor will engage to reduce the gain of an output signal as the input signal (it's actually the sidechain signal) surpasses a set threshold. In doing so, it effectively reduces the dynamic range by bringing down the peak levels of the signal.
The sidechain of a compressor is the signal that controls the gain reduction circuit. It's generally taken from the input signal but can also be a completely different “external” signal. Every compressor has a proper sidechain signal — a lot of confusion, I believe, comes from the technique called “sidechain compression” that makes use of an external sidechain, but whether we use the input or an external signal, every compressor has a sidechain path.
I like to think of compression in terms of the primary goal of mixing, which is balance. In this way, we can imagine a downward compressor as a sort of automatic fader that automatically gets pulled down as the signal gets too high to help balance a track — of course, compression is used for much more than this, but I find it to be a helpful visual.
Upward compression does the opposite by effectively bringing up the output signal level as the input/sidechain drops below a set threshold.
Now let's define the most important parameters available in compressors. I'll be adding links for more information on each of these, and I'll be leaving their definitions (in this article) open-ended, as their definitions change slightly when switching between upward and downward compression.
With that stated, the more important compressor parameters are:
A compressor's threshold is a set amplitude by which the compressor will engage as the sidechain signal crosses it and disengage as it crosses it the other way.
- In downward compression, the compressor will engage to apply gain reduction as the sidechain signal level becomes higher than the threshold (according to the attack time) and disengage as the sidechain signal level becomes lower than the threshold (according to the release time).
- In upward compression, the compressor will engage to apply gain as the sidechain signal level becomes lower than the threshold (according to the attack time), and disengage as the sidechain signal level becomes higher than the threshold (according to the release time).
For more information: Dynamic Range Compression: What Is The Threshold Control?
A compressor's ratio defines how much the compressor will “compress” the audio signal beyond its threshold. It's a ratio that defines the relationship between the level (in decibels) of the sidechain signal beyond the threshold and the level (in decibels) of the output signal beyond the threshold. Therefore, increasing the ratio will increase the amount of compression.
- In downward compression, the ratio defines [sidechain signal dB above the threshold] : [output signal dB above the threshold].
- In upward compression, the ratio defines [sidechain signal dB below the threshold] : [output signal dB below the threshold].
In compressors that offer both downward and upward compression, the switch between them is often in the ratio control, with downward compression having the typical “X:1” notation and the upward compression having parentheses around the first number, noted as “(X):1”.
A 1:1 ratio means that there will be no compression, as the sidechain signal (typically the input signal) will be outputted with no gain change (at least according to the gain reduction circuit of the compressor).
With threshold and ratio defined, we can now take a look at simple input/output graphs to better visualize how compression works.
Let's start with downward compression:
Above, we can see the input level on the x-axis and the output level on the y-axis. There is also a clearly defined vertical threshold.
The solid black line represents the point where the compressor will not be engaged — it's effectively a 1:1 ratio up until we reach the threshold.
Above the threshold, we have our dotted black lines, which represent the input level where the compressor will be engaged. I offer two options here:
- A 2:1 ratio, meaning that for every 2 dB the input signal goes above the threshold, the resulting output signal will only be 1 dB above the threshold.
- A 4:1 ratio (the flatter curve), meaning that now for every 4 dB the input signal goes above the threshold, the resulting output signal will only be 1 dB above the threshold.
So, as an example, let's consider a threshold of -20 dBFS, a ratio of 10:1, and an input signal that periodically and instantaneously switches between -30 and -10 dBFS.
- With the input signal at negative -30 dBFS, the compressor will not engage.
- As the input signal switches to -10 dBFS, the compressor will engage (according to its attack time) and, because of the 10:1 ratio, turn the -10 dBFS input into a -19 dBFS output (10 dB input above the threshold means 1 dB output above the threshold).
- As the input signal switches back to -30 dBFS, the compressor will disengage (according to its release time).
Let's now take what we've learned and apply it to upward compression:
Again, we have the input level on the x-axis and the output level on the y-axis. There is also a clearly defined vertical threshold.
The solid black line represents the point where the compressor will not be engaged, but this time it's above the threshold rather than below it.
Below the threshold, we have our dotted black lines, which represent the input level where the compressor will be engaged. I offer two options once again:
- A (2):1 ratio, meaning that for every 2 dB the input signal goes below the threshold, the resulting output signal will only be 1 dB below the threshold.
- A (4):1 ratio (the flatter curve), meaning that now for every 4 dB the input signal goes below the threshold, the resulting output signal will only be 1 dB below the threshold.
Let's return to our example of a compressor with a threshold of -20 dBFS, and an input signal that periodically and instantaneously switches between -30 and -10 dBFS, only now with a ratio of (10):1 so that we get upward compression.
- With the input signal at negative -30 dBFS, the compressor will engage (according to its attack time) and, because of the (10):1 ratio, turn the -30 dBFS input into a -21 dBFS output (10 dB input below the threshold means 1 dB output below the threshold).
- As the input signal switches to -10 dBFS, the compressor will disengage (according to its release time).
I hope that all makes sense to really drive home the difference between downward and upward compression.
For more information: Dynamic Range Compression: What Is The Ratio Control?
A compressor's attack time is the amount of time it takes for a compressor to engage/react once the sidechain signal crosses the threshold from its 1:1 range into its compression range.
- In downward compression, the attack time refers to how fast the compressor will react as the sidechain signal surpasses the threshold from its lower-level 1:1 range.
- In upward compression, the attack time refers to how fast the compressor will react as the sidechain signal drops below the threshold from its higher-level 1:1 range.
Note that the time parameters in compressors are not delayed action times but rather rates of change. In other words, the attack time represents the time it takes for the compressor to go from no compression to its full compression settings. It is not a period before the compressor begins working once the threshold is surpassed.
A compressor's release time is the amount of time it takes for a compressor to engage/react once the sidechain signal crosses back across the threshold from its compression range to its 1:1 range.
- In downward compression, the release time refers to how fast the compressor will disengage as the sidechain signal drops back down below the threshold from its higher-level compression range.
- In upward compression, the release time refers to how fast the compressor will disengage as the sidechain signal surpasses the threshold into its 1:1 range.
Like the attack time, the release time control is a rate of change and not a delay of action.
For more information: Dynamic Range Compression: Attack & Release Controls
A compressor's knee refers to the transition point around its threshold. A hard knee offers a more distinct triggering of the compressor, while a soft knee allows for a smoother and more gradual transition to compression.
Here's a quick picture to help visualize a hard knee versus a soft knee:
For more information: Dynamic Range Compression: What Is The Knee Control?
Output Or “Makeup” Gain
Most compressors will offer a separate gain stage after their compression/gain reduction circuit to allow users to “make up” for the gain lost (or applied) through compression — this control allows us to maintain the original signal level going into the compressor at the compressor's output.
- In downward compression, the output gain is typically positive, used to make up for the gain reduction applied to the peaks of the audio.
- In upward compression, the output gain is typically negative, used to bring the levels back down from the gain applied to lower-level signals of the audio.
For more information: Dynamic Range Compression: What Is The Makeup Gain Control?
Alright, I hope you're still with me here. Now that we've established the basics, let's move on to the specifics of upward compression.
A Note On Attack And Release Times
If you're familiar with compression, you'll know how important the attack and release times are for dealing in the performance and sound of the compressor.
Fast attack times will cause the compressor to clamp down quickly and reduce the punchiness of our transients. Slow attack times will allow some above-threshold information through before the compress applies its full gain reduction.
Fast release times will cause the compressor to disengage more quickly, allowing for louder results. They can sound more natural with low levels of compression or more jumpy with high levels of compression. Slower release times can smooth out the signal or cause pumping. Very long release times may not allow the compressor to disengage at all, rendering it a glorified volume attenuator more than a snappy, dynamic processor.
Attack and release times work similarly with upward compression, but we do have to think of them in the opposite fashion in terms of the input/sidechain signal level crossing the threshold.
Fast attack times will cause the upward compressor to quickly add gain to the signal as the input drops below the threshold, helping to increase the sustain of notes and create louder outputs. Slow attacks can help smooth things out or produce pumping in more extreme cases.
Fast release times will cause the upward compressor to quickly stop applying gain as the input/sidechain exceeds the threshold, causing the original peaks to remain where they were in terms of level, though causing a perceived “clamping down” relative to the gain applied to the lower-level signal. Slow release times mean that some gain will remain after the input/sidechain exceeds the threshold, leading to an accentuation of the original transients in the input signal.
Use Cases Of Upward Compression
Knowing how upward compression works is great, but how and why should we use it in our music production and mixes?
There are four main reasons why I use (and would suggest using) upward compression, each having to do with bringing up the lower-level nuances of a signal:
- Increase the sustain of certain notes: this can be highly effective on percussion tracks that need a bit of “beefing up” in the mix. It can also work well on held notes of various instruments — I sometimes opt for it on bass guitar to help improve the sustain, particularly in the low end.
- Bring up the ambience of a recording: if I ever want more of the room character (or even some baked-in reverb from a sample) to be present in a track, upward compression is a useful tool for achieving my goal.
- Putting a focus on nuance: the raising of the low signal levels can help bring nuances to the forefront of the mix and, therefore, upward compression can make our mixes a bit more intimate, especially on vocals and instruments with a lot of ghost notes.
- Dirty up a signal: the increasing of the noise floor and worsening of the noise floor is generally considered a bad thing. However, it can be a cool effect in some instances, and upward compression can get us that “effect”.
Upward Compression Vs. Parallel Compression
Although it's not 100% accurate, I often refer to parallel compression as being, effectively, upward compression. There's some truth to that — allow me to explain.
As we know by now, upward compression adds gain to the lower levels of an audio signal relative to the higher levels of an audio signal.
It's important to note that we have our output gain control to either decrease or increase the entirety of the output signal (independent of the compression circuit).
Now, let's imagine parallel compression, a technique where we have a “dry”, uncompressed signal that is also being sent to a “wet” auxiliary track (effects return channel) for heavy compression. The two channels are then mixed together.
What happens in the case of adding the original and parallel channels together? Let's assume, for simplicity's sake, that the uncompressed signal levels of the parallel channel are equal to those lower levels of the dry signal.
So at the lower levels, we have a doubling of the overall level. However, at the higher levels, where the parallel compression channel is applying downward compression to the signal, we have a case where we are getting less than a doubling of level as we mix the two signals together.
In other words, we're adding more gain to the lower levels of an audio signal relative to the higher levels of an audio signal. If we consider adding makeup gain to an upward compressor, we can get the same qualitative result.
And so these two processes are similar and can be used to achieve similar results. However, the way by which they're set up and controlled is different, so we'll treat them as separate yet closely related processes.
An additional benefit of having an independent channel for parallel compression is that we can adjust the parallel channel levels independently (versus a single wet/dry mix control), insert additional processing on the channel, and even send that channel elsewhere in the mixer if need be.
I write about the many advantages of utilizing effects returns in my article Unlock The Creative Potential Of Effects Sends And Returns.
Are Downward And Upward Compression The Same Thing?
Now for the nerdy part. Yes, in an ideal world with the proper use of output/makeup gain, downward and upward compression are mathematically the same.
And yet, they sound different when used in the real world on complex audio signals. So what gives?
Let's start with a theoretical question:
So, downward compression is simply bringing down a signal above a threshold, and upward compression is simply boosting a signal below a threshold. In that case, wouldn't they achieve the same results with the same threshold, opposite ratio, and flipped attack and release times if their makeup gains were set so that their outputs were at the same level?
To investigate the sameness of upward and downward compression, let's do some math and make a few assumptions:
- Assume we're using identical and completely transparent compressors capable of both upward and downward compression.
- Assume there's no attack or release time (for now)
Let's take two identical, non-dynamic sine waves on two separate tracks, for example.
One track (Track A) will have a downward compressor with a 10:1 ratio, and the other (Track B) will have an upward compressor with a (10):1 ratio. Each compressor will have a threshold of -20 dBFS.
Let's set the two identical sine waves to peak at -10 dBFS. What happens to the signals on Track A and B?
Track A: a ratio of 10:1 with a threshold of -20 dBFS means an input of -10 dBFS would output a -19 dBFS level without makeup gain.
Track B: a ratio of (10):1 with a threshold of -20 dBFS, an input of -10 dBFS would output -10 dBFS level without makeup gain
So far, so good. Now, we take into consideration that ratios are linear, so even though we're dealing with logarithmic quantities in our decibels, we can treat the ratios themselves as linear, just like our makeup gain (both using dBFS in this case).
As an aside, decibels are confusing, so if you need more information, I put together this article: What Are Decibels? The Ultimate dB Guide For Audio & Sound.
Moving on, let's adjust the makeup gain of the downward compressor on Track A to match the output of the upward compressor on Track B (a 9 dB boost in this case). Remembering that compression acts on the signal level over time and not instantaneously and that sine waves are very basic (without transients), we now safely have the same sine waves once again at the same level.
Now, let's bring down the identical sine wave levels by identical amounts — let's say 20 dB, so that they both peak at -30 dBFS. What happens to the signals on Track A and B?
Track A: a ratio of 10:1 with a threshold of -20 dBFS means an input of -30 dBFS would output a -30 dBFS level. Adding the 9 dB makeup gain, we get a signal level of -21 dBFS.
Track B: a ratio of (10):1 with a threshold of -20 dBFS, an input of -30 dBFS would output -21 dBFS level without makeup gain
And we once again have our perfectly identical sine waves.
Therefore, in a perfectly sterile, non-dynamic world, the actual gain reduction of upward and downward compression is the same.
But even though the math lines up, upward and downward compression do not behave the same on complex waveforms, even if the attack time of one equals the release time of the other and vice versa.
That is primarily because the complex waveforms that make up our tracks don't pass any arbitrarily set threshold at the same rate — that is, the rate of amplitude change isn't mirrored as the signal goes above any given threshold or below any given threshold.
And, therefore, while the two “equal but opposite” compressors mentioned in this section may work to apply the same compression on a simple sine wave, they won't produce the same results on complex waveforms (a vocal, a snare drum, or, indeed, a mix bus).
Call To Action
I encourage you to take what you've learned from this article and experiment with upward compression in your mixes. Listen critically to how it may differ from downward compression and, even better, from parallel compression.
Find use cases for upward compression that you like, and add them to your toolbox for such instances.
New To Mixing? Check Out My FREE In-Depth Article:
Make the right decisions at the right times to craft professional mixes!
What is the difference between audio compression and limiting? Dynamic range compression and limiting both works on the same principle of reducing an audio signal's dynamic range. Limiting is compression with a very high (often infinite) ratio. Compressors reduce signal levels above a set threshold by a ratio. Limiters are designed to set a maximum output level.
For more information: What Is The Difference Between Audio Compression & Limiting?
What is the difference between feedback and feedforward compression? Feedback compression feeds the audio signal into the sidechain just after the gain reduction element. This compressor type reacts to the signal amplitude without anticipation. Feed-forward compression feeds the audio signal into the sidechain before the gain reduction element. This compressor type anticipates the signal amplitude and adjusts the sidechain signal in advance.
For more information: Feedback Vs. Feed-Forward Dynamic Range Compressors In Audio