First some general stuff that I feel is important context for these knobs- It’s good to approach compression with some caution. When the compressor lowers the signal, you have to boost it back up and this will boost the noise floor from electronics. On modern gear, you typically enjoy a very high signal to noise ratio. Moderate use of one compressor isn’t going to bring this up to a boiling point, but it’s cumulative. Intense use at multiple stages and it will start to creep in. And maybe that’s OK. A lot of use like noise. Intense use with attack and release times that aren’t natural to the source material also introduce artifacts called pumping and breathing. It can be really ugly, or really powerful, or maybe you want really ugly. Play around, have fun, take control! Again, you’re more likely going to be unhappy with your results if you end up with a lot of small, unintended artifacts all over your track.
Essentially, you feed something into the detector circuit not unlike control voltage, and a gain element like a VCA turns the signal down automatically based on your preferred timing, ratio, curve, and sometimes other things. By default, if you don’t change anything, it’s safe to assume that what you’re feeding into the detector circuit is the same signal/track/sound that you’re compressing. The original most common use case was taking a musical performance, spoken word, etc. with inconsistent dynamics, and making them more consistent. You’re making a punk backing vocal track and tried to sing “OI! OI! OI!” but you had a frog in your throat and you sang “OI! oi. OI!”. You set the compressor to turn 1 and 3 down a bit then boost 2 and you’ve got something consistently loud but it still has some human vibe.
In electronic music, we don’t deal with that very much. We go the other way and inject some randomness and humanity into things that are perfectly consistent and boring by default. When we work with samples from a library, device, expansion, etc., assume it has already been compressed to some degree (and the signal to noise ratio already lowered). The only exception will be things you sample yourself or material that’s very explicitly marked as such. But sometimes we’ll have a mix with dynamics all over the place and just want everything to feel a bit more cohesive AKA “glue”. Some basic compression at moderate attack and release times can be great for this. But say you’re working on an electronic track with a prominent kick and it’s triggering that detector circuit harder than any of the stuff you want to actually blend together.
Earlier I alluded to the fact that you send something into the detector circuit and by default it’s just the signal you’re processing but compressors sometimes have ways you can manipulate the signal before it reaches the detector, such as a high pass filter to curb some of that kick energy so it can go to work on the rest of the tracks. Sometimes you’ll get a dedicated side chain input where you can control the compression with a totally separate signal. ie you’re making a club track and you love that the compressor is going nuts every time the kick hits so you feed just the kick track into that detector circuit and your track just pumps like crazy. It can even be a separate, modified kick track that no one hears with special processing to tailor that pumping to your taste.
The trickiest part of compression is that much like EQ, the visual representations we have are impact and only tell part of the story. Like if you told a very basic artificial intelligence to drive the car the best it could, it would just red-line the engine until it blew. A better understanding of the detector circuit and the frequencies you’re feeding into it on top of just seeing the gain reduction is ultimately how you take total control of a tool like this. In the analog world, we’re generally referring to the detector circuit when talking about types of compressors. VCA compressor (SSL G bus comp, etc) are kind of like on a synthesizer but with a track controlling it as opposed to an envelope. Or in the case of sidechaining a kick track, a lot like an envelope. It’s know for sounding remarkably natural and transparent with natural settings, but can do some very cool pumping as well. The compressor in the Digitakt is closest to a VCA type, not necessarily modeling SSL or “The Glue” behavior, but you’re in that ballpark and you can totally use those settings. It’s pretty safe to say AH+ is too, as well as most generic digital built-in compressors unless otherwise specified. Opto (LA2A) use vactrol/photo cell circuits like in a low pass gate. Unique sound, very cool for vocals. Slower time settings can start to get in that ballpark. FET compressors are, big surprise, based on more basic transistors and are capable of being crazy fast and aggressive.
If you use side chain functions or EQ before compression and really use your ears, you’ll start hearing how these circuits bite down differently on high frequencies vs low frequencies. Our hearing doesn’t work the same way as these circuits. Volume changes on a logarithmic scale. We have the Fletcher-Munson curve which shows that we don’t hear all frequencies equally. Probably something to do with lower frequencies being associated with larger animals that would kill us and other cool evolutionary monkey-brain stuff. Well, we went ahead and used logarithmic controls for manipulating gain and we’re only ever gonna hear it the way we hear it whether it’s coming in or going out so this almost becomes a moot point. For many applications, a compressor is just turning the volume down and bringing it back based on how we would hear things. But once we embrace that these incredible tools can do stuff we can’t, like react specifically to certain ranges of frequencies faster than we can perceive, you start getting into the kind of voodoo that makes the people who have mixed some of your favorite tracks so extraordinary.
You can take pretty direct control over the strength of the attack and the decay of prerecorded audio. Transient shapers are, at their core, compressors with some simplified controls and cool sidechain stuff behind the scenes. You can severely reduce the resonance of a drum while preserving the hit and vice versa. At intense settings, you can make significant changes to a groove in terms of timing and intensity and because it’s Elektron, you can change this over the course of a section, song, dynamically, with the mod matrix.
Best I can tell, AH+ does not let you route a totally separate track to the sidechain like the Digitakt and some other compressors do. You can target every parameter on the compressor with an external signal at the CV input and target the mix ratio or the threshold so that it only starts working when another track plays. Where many compressors only have a high pass side chain filter, AH+ gives you low pass as well, so you can get that big pumping effect going without being able to route a dedicated kick track. Even that thing is on the mod matrix so you could be gluing tracks with the high pass during one section then pumping like crazy for another with the low pass. You could take pitch CV and do key-tracking compression.
Just like the former AH, you can approximate a compressor using the trick in the manual modulating gain with the envelope follower. This is not only a great way to understand compression by building it from other tools, but of course you get to bring saturation and distortion into the mix, as well as run 2 compressors in series with both doing half the work, different jobs, whatever you want.
