Multi-stage Compression

Photo: Universal Audio

When and why might you want to reach for more than one compressor to process the same signal?

Every engineer needs to know how to use compressors to manipulate an audio signal’s dynamics. Often, you’ll be able to access the control you want with a single compressor, and you should certainly start by figuring out what you can achieve that way (check out Sam Inglis’ SOS November 2019 article if you need advice on how to operate a compressor. As you clock up more mixing miles, though, you’ll increasingly encounter situations where no one setting seems to work for the whole duration of the song. It’s often possible to solve such problems by using two compressors, and there are also some interesting mix strategies which employ multiple compressors. I’ll explore a number of such techniques in this article.

First, it’s worth pointing out that many of us already use multi‑compressor setups without giving it a second thought. If you use mix‑bus and subgroup compression (eg. on the drum bus), then plenty of your sources will flow through at least three compressors: one for the source, one at the group bus, and one at the stereo bus. Some of us use compression while recording too, and other processors might compress the signal as part of their sound; amps and cabinets, tape saturation and distortion effects all do that. And lots of us use a compressor and a limiter on the stereo bus or as part of a DIY mastering chain: a compressor to ‘glue’ the mix together, and perhaps give it a little more ‘punch’ or ‘solidity’, and a limiter (which is simply a high‑ratio compressor) to rein in peaks, preventing ‘digital overs’ and perhaps buy headroom to allow the signal level to be raised.

One To Another

With all that in mind, let’s start with another compressor‑plus‑limiter technique. A time‑honoured tactic is to use a UREI 1176 FET compressor and an Teletronix LA‑2A tube optical compressor in series on a vocal. These analogue compressors have different characteristics. The 1176 can be set to act very fast, its transformers add some character, and it’s pretty versatile too, with a choice of ratios and adjustable attack and release times. The LA‑2A can sound lovely on vocals: smooth and ‘natural’, with a slight thickening effect from the transformers and a pleasant, subtle euphony from the tube amplification. But it’s far less controllable than the 1176; you can switch the ratio between Limit and Compress, but most of its behaviour is automatic, varying with the complexity and level of the signal. So, while an LA‑2A in compression mode can sound beautiful on a vocal, it’s not uncommon that, having set it up to sound its best for most of the track, you later notice points at which a higher signal level triggers too much gain reduction and makes the release a little too sluggish.

A classic vocal combo: the slower‑acting LA‑2A optical compressor can often perform its desired role more easily when preceded by a faster compressor such as the 1176.

In this classic vocal compression chain, you’d place an 1176 before the LA‑2A, with the 1176 set with a fairly fast action and high‑ish ratio (it could be anywhere from 8:1 to the all‑buttons‑in mode, depending on the style) to bring those problematic peaks down by, say, 3‑4 dB. This allows the LA‑2A to perform more consistently. By the way, there’s no reason you can’t use this chain on other sources, or use these compressors the opposite way around (the LA‑2A set to Limit and the 1176 providing the less aggressive compression). You could use different compressors entirely if you wished — various successful producers have suggested combining the 1176 or another fast compressor (ie. FET, VCA or digital) with a Fairchild 670, RCA BA6A or UA 175 (all variable‑mu types and slow in comparison with FET and VCA ones) in similar fashion. But a helpful guideline is to make sure the first compressor is the faster‑acting one, and that it has a higher threshold than the second.

This isn’t the only reason you might chain two compressors in series, of course. By using two (or even three!) compressors with different attack and release times, you can often target the gain reduction of each processor at specific qualities in a sound. For example, you might have one focusing on the note onsets of an acoustic guitar, and another with a slower attack that aims to control the ‘bloom’ in a longer note when a string is picked that bit too hard. That said, to my mind, multiband compression or better still dynamic EQ (itself technically a form of compressor!) is a better tool for that sort of job.

Sidechain Tactics

If you want to make the main compressor’s job easier but don’t want to process the sound as in the 1176/LA‑2A serial compression approach, an alternative option is to compress the sidechain control signal.

Though the 1176 and LA‑2A combo can work well, the first compressor does have an impact on the sound. If you’d like to prevent the LA‑2A getting too overenthusiastic with the gain reduction without changing the louder parts, you can set it up to respond to an external sidechain signal, and insert your faster compressor on this sidechain signal. Most DAW software makes this easy to set up: just create a mult (duplicate track) of the part you’re processing, route the mult to the vocal compressor’s sidechain input (being sure not to also route it to the stereo mix bus!), and then place your faster compressor on this mult. This will stop the main vocal compressor from overreacting to the louder signals, but those signals themselves will pass through unscathed.

While we’re on the subject of sidechain processing, a far more common tactic is to EQ the sidechain signal, for example using a high‑pass filter to reduce a compressor’s sensitivity to low frequencies, or an API’s 2500’s Thrust control to increase the sensitivity higher up the frequency spectrum. You can have far more precise control over the compressor’s sensitivity to different frequencies, though, if you use a dedicated EQ plug‑in on a sidechain control track (a mult of the source, as described above). And this makes it possible to target multiple compressors, inserted in series, very precisely using band‑pass filters on the sidechain. For instance, if you’re hoping to rebalance a stereo drum loop, you can fairly easily make one compressor react only to the kick drum, and make another more sensitive to the snare or the metalwork. A word of caution though: since you need a different control signal for each compressor, this tactic can be fiddly and time‑consuming, so it’s best to avoid this trick until you actually need it!

Parallel & Series

A simplified routing setup for a dual‑stage parallel compression technique that can be useful when the dynamic range changes significantly in different parts of the song. The drums are sent post‑fader to a compressor on an aux bus, and both the drums and this aux bus are routed to a stereo bus, on which the main drum compressor sits. When the song is quieter, the parallel compressor makes a significant contribution; when it’s louder it’s dominated by the dry signal, and the main compressor takes over.The rest of the strategies I’ll describe here all use parallel compression, and the first also employs two compressors in series. It can be used when compressing a signal whose levels and dynamics change significantly from one song section to another, and can be particularly helpful on a drum mix in songs with lower‑level verses and louder, higher‑energy choruses. But it can work well on other sources too and probably (I’ve not yet had to try it) on the stereo mix bus; anything where the dynamics change significantly from one song section to another. The conundrum we’re trying to decode is this: if you set your main compressor’s threshold low enough that it does what you want in the quieter sections, it will suck the life out of the more energetic passages that are supposed to punch harder; but if it’s optimised for those louder ‘payload’ sections, you’ll have no control over the calmer moments.

There’s more than one answer to this riddle, but this particular technique starts with a parallel compression setup, and places a regular compressor after it. You start by setting the second compressor to do what you want during the louder parts of the mix/performance: making the big‑hitting parts, erm... hit big! Then turn to the parallel compressor to target the quieter sections. During these, the compressor’s ‘wet’ contribution to the wet/dry blend is significant, so you can use this tactic to bring up lower‑level details to thicken the sound in this section if desired. During the louder passages, on the other hand, the unprocessed (dry) sound will dominate that mix, so the parallel compressor doesn’t make such a noticeable contribution. Thus, the second compressor still does what you want, exactly when you want it to.

By the way, I recommend performing parallel compression using your DAW’s routing facilities rather than relying only on plug‑ins with a wet/dry control. For one thing, it makes it easier to see on your mixer channels what’s going on, but it also allows you much greater control over the parallel compressor’s contribution, since you can EQ the wet signal (something I pretty much always do when using parallel compression).

Joe Chiccarelli: Vocal Tone

Joe Chiccarelli at Sunset Sound.Photo: Ana GibertThe next three techniques have all been made famous by A‑list mix engineers, and we’ll start with Joe Chiccarelli’s approach to vocal compression. If you’ve already read our review of IK Multimedia’s Joe Chiccarelli Vocal Strip plug‑in (elsewhere in this issue), you’ll know that Joe designed this plug‑in with IK, and that it places three compressors in parallel with the dry sound. There’s more to his chain, including reverb, delay, saturation and modulation, but he says this compression setup lies right at the heart of his approach to vocal processing, particularly in big rock mixes where vocals sometimes struggle to remain heard among the thick guitars.

The idea is that different compressors are chosen for their sonic character as well their gain‑reduction characteristics, and each suits different song styles and vocalists. Having these three options patched in and ready to go makes it quick and easy to craft a sound that works, and he can change the balance between the four signal paths in different song sections where necessary. While the plug‑in makes setting up this chain really quick and easy, you can set up something similar in any DAW and, if you wish, experiment with different compressors. Joe uses a stereo pair of 1176 FET compressors, a Fairchild (a vari‑mu compressor), and a UA 176 (a vari‑mu type which pre‑dates the 1176) — although in the IK plug‑in, this last one is a hybrid model based on the UA 176 and tweaked, after experimentation, to incorporate some aspects of a Tube‑Tech CL 1B (a tube‑amplified optical compressor). The FET compressor is intended for a “more natural” sound, the Fairchild for a warmer tone, and the last of the three for a heavier, rock‑style sound, and set to compress more assertively than the others.

Andrew Scheps: Rear Bus Compression

Next up, we have Andrew Scheps’ ‘rear bus compression’ technique, so called because he first set it up using the ‘rear bus’ of his Neve 8068 quad console. All tracks are routed through the stereo bus as normal, but some are also sent in parallel to this rear bus, on which a compressor is inserted. Now working in the box, Scheps uses a multi‑mono (ie. unlinked stereo) 1176 modelling plug‑in, set with a low ratio, slow attack and fast release.

Andrew Scheps’ ‘rear bus’ compression technique, which can be a useful tactic for energetic‑sounding mixes, is named after the rear bus on a Neve 8068 quadraphonic console, and can be used in conjunction with regular stereo mix‑bus compression.Importantly, not all tracks are sent to this bus. It seems that what Scheps sends there has changed over the years, so feel free to experiment, but last year, he said he now sends “basically... everything in the mix except the drums” to this rear bus compressor. The idea is to use compression to glue all these sources together and make them interact, with the lead sounds tending to trigger most compression (unlike with regular mix‑bus compression, where the drums tend to do that), and the overall effect is to make the mix sound punchy, energetic and exciting. It can be very effective, particularly on rock/grunge sorts of styles featuring distorted guitars.

What’s that I hear you say? “But this only uses one compressor!” Well spotted. Part of the beauty of this setup is that you’re still free to process the stereo mix bus in the usual way, and I’ve found that a touch of compression there, just tickling the meters, can really help the drums and everything else gel — you don’t have to push the gain reduction too far in search of excitement, since the rear bus is already bringing that to the party.

Brauerizing: Multi‑bus & Parallel Compression

We could explore more famous names here, but I have to stop somewhere. Still, no discussion of multi‑compressor techniques would be even half‑complete if it ignored Michael Brauer! His famous Brauerizing approach is more sophisticated and nuanced than I’ve seen some online tutorials suggest — often they stop at the idea of multi‑bus compression, but that’s only the first chapter of a story which also involves a dedicated multi‑parallel compression setup for vocals, and what Brauer describes as ‘send/return’ channel compression. If you want to dig deeper into Brauerizing, it’s well worth reading through the Q&A on his website (www.mbrauer.com/q-a), as he offers some great insights and clarifies various misconceptions about the technique. With that in mind, I’ll offer a brief overview of the setup, as I understand it.

Michael Brauer has developed a sophisticated parallel processing approach for his mixes — there’s much more to it than the widely reported ABCD multi‑bus compression system.Brauer mixes, from the outset, ‘into’ five separate buses: four (A, B, C and D) with a compressor inserted (sometimes with other processing too), and one clean, with no processing. These five buses are summed to create the stereo mix. The compressors’ Left and Right channels have the same settings but, as in Scheps’ system, they’re unlinked so that they react only to ‘their side’ of the stereo signal. The compressors’ thresholds and output levels are carefully calibrated with a test tone, and specific groups of instruments are routed directly to one bus: drums and bass go to one, guitars to another, and so on. Brauer says each compressor typically applies no more than a couple of dB of gain reduction. The unprocessed bus can be used for pre‑compressed sounds (all the sounds in his mix projects will probably already have had some compression applied, but things like samples and loops might be more heavily processed). This technique also allows him to send sounds to the unprocessed bus to bring them up in level without interfering with what the compressors are already doing.

Brauer says he uses this multi‑compressor setup to create a sense of energy, excitement and urgency, and to my mind it makes this possible in several ways:

• First, as with the Scheps rear bus approach, it allows the engineer to mix ‘into’ compression without some sources triggering unhelpful gain reduction in others.
• Second, it allows compressors and settings to be chosen specifically to suit each group of sources, with different attack/release times and curves, different harmonic distortion contributions, and perhaps ‘special features’ (such as the stereo width control on Brauer’s TF Pro P8 Edward The Compressor).
• Third, because the bus compressors are all different types and processing different signals, they react in different ways at different times — some sounds being pushed while others are pulled, and this can lend a sense of life and urgency to the mix; the unlinked L‑R channels can do a similar thing when it comes to stereo width.
• Fourth, mixing into the compressors can help nudge you more quickly towards a pleasing mix balance: if you notice too much compression being triggered, it’s an indication that you should try turning down the loudest signal being sent to that bus, which can be easier to identify in a busy mix than when everything runs through the same compressor. And if working in the analogue world, it’s a studio assistant’s dream, since all the compressors have the same settings for pretty much every mix: once you’ve recalled the console your job’s almost done!

We’ve already looked, in the Joe Chiccarelli section, at the idea of using multiple parallel compressors for the vocal, so I won’t dwell on that aspect of Brauerizing here, but Brauer has stressed the importance to his approach of what he describes as ‘send/return’ compression. This is a variation on parallel compression which is intended, as he puts it, “to add tone, fatness, attitude and urgency to a sound... it’s where the emotion starts to really develop.” Again, this is all about mixing the various sources into compression, rather than start by controlling individual sources with their own compressor. To make this possible, compressors are set up on six aux buses (one stereo, and five mono), which ‘return’ to spare console channels. As with his ABCD multi‑bus setup, each compressor is chosen for its specific characteristics. Then, as he’s mixing into the buses, if he feels a sound needs more attitude, thickness or a tonal change of some sort, he can use the channel send controls to feed some of the source to the appropriate compressor, almost as an alternative to EQ. The stereo send, with two unlinked 1176s, seems a particularly important feature for him — he goes into some detail about how to set these up in his aforementioned online Q&A if you want to learn more about that.