When trying to understand compressors, it helps to know the different behavior of 1176, Tube, VCA, Optical, Digital, and upward compressors. By knowing the parameters of these different compressor types, we’ll know which ones cause distortion, which work well for particular instruments, and how they affect the frequency response.
Before you even begin to compress your signal, a compressor may be affecting your audio by subtly altering its frequency response - for example, a FET emulation may subtly attenuate the highs, or a tube compressor may dip lows and highs. This all depends on how the plugin was designed.
If we look at the FabFilter Pro-C 2, we see that even when emulating an optical compressor, or a bus compressor that emulates a VCA type compressor, the frequency response isn’t affected - instead only the compression behavior is altered.
Let’s listen to some different compressor types without any actual attenuation occurring to see if we can hear the subtle differences in the frequency responses.
Different compressor emulations will have different points at which they begin to distort your signal. For example if we run a 100Hz sine wave through a FET compressor, we’ll notice subtle 3rd and 5th ordered harmonics, which become more aggressive with a faster attack and release time.
On a digital compressor or non-emulation compressor like the Pro-C 2, we get moderate harmonics with the clean option, aggressive harmonics with the Punch style, and no harmonics with the Mastering Style.
Keep in mind that how much you compress will also affect the harmonic’s respective amplitudes and formation.
Now that we understand some points about all compressors, let’s look at each one in detail - FET compressors utilize transistors for their compression, but since we’re covering emulators in this video, let’s focus more on functionality. This compressor’s attack ranges from 20 microseconds to 800 microseconds.
The release ranges from 50ms to 1.1 seconds. We get 5 ratio settings, 4:1, 8:1, 12:1, 20:1, and an all button which is program dependent and will range from 12:1 to 20:1.
This version also offers a side-chain section in which you can determine if you want the lows to trigger compression, the highs to trigger compression, how the incoming signal is detected, and more.
Due to this compressor's quick attack, it’s always going to slightly distort the signal, meaning it emphasizes transients. This makes it great for drum overheads, parallel compression, and anything you want to give an aggressive sound.
Let’s like at an emulation of the Retro STA since it’s a popular one - the ratio is more or less fixed at a non-linear 3:1, but this is variable depending on the level of the incoming signal. The release ranges from 2 to 8 seconds.
That said, both the attack and release can be sped up by increasing the mode value, but no matter what, both stay pretty slow and smooth sounding. At single mode you’ll the attack is too slow to respond to transients, allowing for some transient retention.
This compressor is almost like the opposite of the 1176, in that it’ll reduce detail and punch, instead causing smooth compression.
VCA Compressors are where you begin to see the threshold become controllable - additionally, this compressor uses RMS detection instead of peak, meaning the threshold and detection are measuring the average loudness, not the loudest point of the waveform. This causes compression that closely emulates how the human ear compresses sound.
But know that not all VCA compressors use RMS detection.
Since the DBX 165 uses a soft knee, our threshold goes above 0dB, but the curve can still very gradually cause attenuation. Our attack and release times can range pretty drastically and can result in both a smooth or punchy sound.
The LA-2A limiting amplifier has a fixed attack of 10ms, but the photocell it uses (or in this case emulates) gives it a unique program-dependent release. The first half of the attenuated signal is released after 60ms, but the other is released between 1 and 15 seconds later.
How long it takes for this other half to be released depends on the frequency and amplitude of the detected signal.
This unique character makes it great for vocals and bass, since you retain some transient detail, but get a fuller sound due to the extended-release of the second half of the attenuated signal.
Modern digital compressors are essentially a combination of all of the compressor settings and technologies of the past - they let you mix and match all of the different functions we covered, as well as add some new ones. They’re by far the most versatile compressor plugins to use.
With them, we can quickly emulate multiple compression attack and release types, and control how much harmonic distortion occurs. Additional functions like lookahead allow the compressor to measure the incoming signal prior to processing it, causing more transparent compression.
We often get a side chain section which allows us to determine what triggers the compression.
A newer form of compression that I like to use a lot, but doesn’t get discussed too often is upward or low-level compression - instead of compressing from the peaks down, it compresses from the noise-floor up. This means we don’t affect our peaks or cause harmonic generation.
Instead, we increase the quietest details of the signal. I like this type of compression a lot since it’s a great way to increase detail but doesn’t reduce transients like most compressors.
The only downside is that it will increase the level of your signal’s noise floor.
Now that we understand the different compressor types, let’s look at how and why the attack and release functions on downward compressors cause or avoid distortion. As we covered earlier, shorter attack and release times increase the amplitude and sometimes number of harmonics, while longer times do the opposite.
The reason being, at short enough attack and release times, the compressor is working faster than the full length of the waveform.
Let’s explain how this happens. Since 1Hz = 1 full oscillation per second, we can say that a 50Hz wave has 50 oscillations in 1 second.
Let’s divide 1000ms or 1 second by these 50 oscillations to see how long 1 oscillation takes. 1000/50 = 20ms per oscillation.
With that said, if our attack or release is shorter than 20ms, it will likely distort any waveforms 50Hz or higher.
That’s why the 1176, with its 20 to 800 microsecond attack time almost always causes distortion.
Last up let’s talk about how different plugins as designed with different levels of quality - let’s look at Logic’s stock plugin and observe the harmonics it forms. We notice that we get an aggressive amount of harmonics with just about every compressor emulation type - with maybe the exception of platinum digital.
But if we switch to Studio VCA, we see a lot of high-ordered harmonics, which often have a harsh and unwanted sound. We even see comb filtering caused by phased cancellation.
What’s causing this is aliasing distortion. Since the compressor wasn’t designed too well, it distorts high frequencies. Without oversampling, which the plugin doesn’t introduce, harmonics generated from these distorted high frequencies have nowhere to go and are reflected back down the frequency spectrum.
What’s worse is that these harmonics can be disharmonious with the key of our song. So in short, be sure to use a compressor that uses oversampling and filtering to avoid this issue.