Here’s a behind-the-scenes look at what we do here on the iZotope blog: for the article I wrote on common compression mistakes, I had originally listed the following tip:
“When many engineers say ‘compression’, what they mean is “downward compression.” In other words, bringing down the level of the signal above the threshold that you set on your compressor, to make louder things quieter. But all too often, we forget about upward compression, where quieter sounds are brought up to the threshold point; this technique can be quite handy in certain situations for a more transparent effect (it can also be approximated with parallel compression, if you don’t have an upward-compressor on hand).
Similarly, expansion—of both the upward and downward variety—can greatly help your mix. If I’m given a dull snare sound, I might not reach for a static EQ, but for upward expansion. That way, the harder the drummer hits, the more attack, snap, and other goodies I can bring out, leaving nasty resonances alone for the ghost notes and softer strokes.
Let’s take a look at three other dynamic processes often overlooked in mixing and production, with tips on how to employ them.
To quickly recap, in what we commonly call “compression,” a signal is brought down in level when it crosses a dialed-in threshold. More accurately, this is termed “downward compression.”
Upward compression, however, works from the opposite end of the spectrum: when a quiet signal falls below a predetermined threshold, it’s brought up in level.
If you have a working knowledge of expansion (covered later, don’t worry), you’d be forgiven for scratching your head. Surely this must be expansion, right? After all, you’re pushing a signal up, not pulling it down.
To that point, I’d reply that you must take into account the governing context of the signal as a whole: In upward compression, the overall dynamic range of the material is still being reduced, not expanded. Therein lies the difference. Yes, the lower level is now louder, but the higher level remains the same, and the net effect? The dynamic range has been compacted—just like in regular old compression.
I like to use an upward compressor on any room-based track that isn’t quite roomy enough. Let’s take the room mics of a drum set: in the studio, perhaps the drum booth was rather small, resulting in a closed-in sound. Using an upward compressor in this scenario can allow you to bring out the ambiance—the space between the transients—and get a roomier feel in your track.
You might be saying, why not use a transient-shaper for this? You sure could—but they aren’t very transparent as a rule; they can often add distortion. Sometimes I love that distortion. But sometimes it’s uncalled for, and that’s when an upward compressor is your friend.
You can also use an upward compressor to create a room-mic sound when no room mics were actually used in the session. In fact, recently I did this on a mix, turning a spare tom mic into ambiance.
Let me explain: I was mixing a live track for the indie band Leland Sundries, and I wanted more room tone than was provided in the recording. I didn’t like what artificial reverbs were giving me, as this was a tightly miked, live-mix scenario. Luckily for me, the rack tom on this tune was never played in the song—not even once! Instead, the mic had picked up a weirdly balanced picture of the whole kit: kick, snare, hat, cymbals, and floor tom were all represented in excellent proportions.
But it wasn’t roomy sounding. Squashing the track with a downward compressor wouldn’t have worked, because that would’ve emphasized the transients, and I was going for the space between the transients. With an upward compressor, I was able to bring out the ring of the kick and snare, turn up the splash of the cymbals, and emphasize other room-based reverberations.
If you want to try this for yourself, it’s possible with many popular software compressors. The compressor in Neutron 2 can be toggled into upward mode when you use negative ratios.
Other plug-in manufacturers supply compressors that work in an upward fashion too; if you see a range value in your compressor, and the range control can be oriented in both a positive and negative directions around the zero point, then chances are the processor can provide this variant on compression.
Just make sure, when experimenting, to pay extra-careful attention to your threshold, attack, and release parameters, and make sure to use very low ratios (otherwise it’s hard to control). To successfully pull off upward compression requires an exacting touch, because the way transients react to this process is somewhat counterintuitive—i.e., you’ll probably want a faster attack than you’d think, and the release will behave differently as well. I tend to use fast attack values and fast to medium fast release settings when utilizing upward compression.
In drum room-mic scenarios, multiband upward compression can also be your friend, allowing you to avoid affecting the cymbals too much, and letting you dig into the snare and toms. Your mileage may vary though—in fact, you might find yourself with no upward compressor in your arsenal at all...
...Or so you would think, if you didn’t read this tutorial!
It turns out that you can use parallel compression to approximate much of what upward compression can give you. It works as follows:
Send an instrument or stereo bus to a compressor on aux track. Set the threshold very low, so that the compressor is constantly working, giving you gain-reduction on soft passages and loud passages alike. Counter to how you might use a compressor on the track itself, you’d employ very fast attack times, and, if you’re like me—that is, generally preferring faster releases for musical pump—you might want to employ slower releases than you’d normally use; the key here is to juice the release settings so they won’t introduce distortion.
These settings, auditioned in solo, might not sound very musical, but when you blend the parallel compression back in with the unaffected signal, you’ll start to feel how it works: because the compressor is kicked into high gear on loud moments, you won’t really notice the parallel track when the music is really hot; however, when the music dips down in level, the parallel track will be more audible, and bolster the overall level of the material. The result? Lower levels have been brought up, almost as they are in upward compression.
Upward expansion is akin to regular old compression in a key way: it affects a louder signal as it crosses the threshold. However, instead of clamping down on the signal, the expander pushes the sound up in level. Here is where the process earns its name—it expands the overall dynamic range of the signal.
By exactly how much is determined, in part, by the ratio. Ratios in compression are expressed in whole numbers greater than one (i.e, 2:1), as you very well know. In expansion, though, the top of the ratio is usually expressed as a fraction of one. So if your compressor can give you a ratio of 0.5:1, then most likely it’s expanding at that point.
We could go deeper into the math of why this is so, but let’s cover best use-cases instead. A good candidate for upward expansion would be a severely-compressed snare track that seems devoid of life—something you received from an inferior recording session. Using an upward expander, you can, to some extent, recreate a sense that the snare is jumping out at you.
The trick is to edge the threshold into a moment when a thwack of the snare is just a wee bit louder than the surrounding sonic information (the rest of the drum leakage, in other words). When you find this sweet spot, the attack, release, and ratio settings will help you shape and boost that initial thwack. Shorter attack times will grab the hit sooner, and quicker release times will bring it back down to its original volume more quickly. The opposite, of course, is true. Electric basses that have been squashed during inferior recording sessions can also benefit from expansion employed in this way.
One can also use expansion’s attack and release parameters to grab unexpected parts of the signal and raise those, thereby creating an unnatural/off-kilter effect for aesthetic purposes. Imagine if a dramatic tom hit rose in level not on the thwack of the tom, but on the resulting ringing. That would be weird…but it could be awesome! You could achieve this effect with a slower attack, because the expander would still be reacting to the initial transient. It would just react far later than you’d expect. This kind of trickery might work with experimental rock of the sort Deerhoof produces.
It’s also worth noting that upward expanders can be employed creatively on sampled material to add a bounce that simply wasn’t present before. All you need to do is find an element within that sample, center the threshold around that element, and adjust the other parameters to bring it out.
Personally, I often use upward expansion in multiband. Bringing out the body of a snare drum just on the hard hits can be achieved handily with an upward expander looped into the 100–220 Hz range. You can also use this technique to bring out the warmth of a thin guitar with too much pick attack, particularly if you sidechain the expander to the guitar’s most brittle frequencies. That way, when the excessive picking noise comes in, the attack brings up the pleasant midrange at almost the same time.
The possibilities for implementation are many. Just be aware that expansion often behaves oppositely to traditional compression. For example, where you would use make-up gain to raise the level of compressed instrument, you might need to attenuate the level of an expanded sound, so it doesn’t jump in gain. Just make sure to create headroom before this in the signal chain in order to avoid clipping.
Downward expansion (conventionally referred to as just “expansion”) is similar to its upward counterpart, but with one fundamental difference: when a signal drops below a set threshold, the downward expander pulls it even farther down toward the noise floor. You probably use downward expansion quite often, though you might not know it—it’s the very process that powers a conventional noise-gate.
Still, downward expansion can be used for far more than gating. Rather than kill extraneous sounds outright, you can use downward expansion as a subtler, more transparent transient-shaper. This is not unlike upward compression, which can also be employed like a transient shaper. Just think of it this way: an upward compressor is similar to pushing the sustain parameter up on a transient-shaper, while a downward expander is not unlike pulling the sustain parameter down.
On drum samples this can come in particularly handy. Say you’ve got a loop with an awesome feel, but you want to emphasize its transients and downplay the atmosphere. You can use downward expansion to shape the decay of the ambiance, thus emphasizing transients in a subtractive process. This can sound audibly less distorted than using a transient-shaping plug-in—or a traditional noise-gate—to achieve the same effect. It can also sound very distorted if you use quick release times (which can be fun).
Downward expansion, like its upward variant, also works well in multiband settings. If you’ve got kick or high-hat leakage on your snare, multiband downward expansion can help take it out of the mix. Simply put the processor on your lowest frequencies for the kick, or on your highest frequencies for the hat. It won’t mitigate the problem completely, but you’ll be able to tame it. You can also use expanders on 808s to duck them out of the way of other bass information quickly. This is useful in hip-hop.
Here’s another multiband trick: you can use a multiband downward expander like a dynamic EQ, though of course, it’ll act counterintuitively, due to a fundamental difference in operation; whereas a dynamic EQ will attenuate frequencies when they hit harder, a multiband downward expander will cut frequencies when they’re decidedly quieter.
Why would we use such a technique? Let’s say you’ve got a bus with a whole bunch of guitars chugging along. In the 200–400 Hz region they sound awesome, but only when the song kicks into full gear. When these guitars sound more softly, their thickness gets in the way of the rest of the mix, maybe because the drummer is playing quieter, and the snare isn’t cutting through as much.
This is a case for using a multiband downward expander over a dynamic EQ: when the guitars hit hard, the chunkiness is heard in all its glory, but in softer moments, those chunky frequencies are more tame. Of course, don’t be too drastic. All you need to do is shave off a couple dB or so.
Compression, in its classic form, is a beautiful thing—but it isn’t the only thing. And, it’s worth noting that traditional compression techniques have been employed for so long that even some EDM producers are getting sick of clamping down on signals. What have they turned to instead? According to Andrew Eisele—whom I interviewed for a piece on mixing electronic music—the answer is expansion.
Yes, tides turn, and this tide might be turning…upward. So, when it comes to dynamic processes for mixing purposes, it is our sincere hope at iZotope that we’ve, ahem, expanded your knowledge.