Mix Rescue: Pocket Lips

Fortunately, the band had been able to keep decent separation between the recorded parts while they were tracking, and spill and phase relationships imposed very few editing restrictions — so I exercised the audio scissors to tighten things up a bit. When working with good live drumming like this, quantising everything to your sequencer's grid will kill the groove stone-dead, so I left the drums well alone. Instead, I used them as a reference point for editing the timing of many of the other parts, starting with the bass and then slotting in the turntables, keyboards and vocals on top, one at a time.

I dealt with techniques for editing timing in quite a bit of detail for SOS October 2008's Mix Rescue, so I won't go into all that again here. What I will add, though, is that this editing accounted for about half the overall mixing time, and I tried to get all of it out of the way before getting into any serious mixing decisions: editing and mixing are mentally very different tasks, and it's easy to get bogged down in one to the detriment of the other when you try to do them at the same time.

While I was at it, I took the opportunity to mute various things so that getting an upfront sound would be easier. Let me explain how this works. In most busy mixes, the listener's ear is constantly 'refocusing' on different parts as they all jockey for attention. To take a visual analogy, it's a bit like watching a football match. While you're focusing on the player in possession, the other players are only in your peripheral vision — and are seen in less detail. When the ball's passed to another player, you shift your focus to him instead, and it's then the first player's turn to be in your peripheral vision.

Where this analogy breaks down, though, is that audio parts to which you're not specifically paying attention will still, to some extent, obscure the sound your ear is focused on, making that less clearly audible than it might be. If you can switch off some of the competing parts while they're psychologically in the background like this, you'll usually find that the foreground stuff sounds more upfront. What's more, non-technical listeners rarely notice if you briefly drop out some background parts — because their attention is, of course, directed at the sounds at the front of the mix.

But what did I actually do here in practice? Well, there were several features I wanted to highlight. The first was the characteristic and hooky double snare-hits at the ends of phrases in the drum part, so I dropped out the bass and synth parts to make these jump a little bit forward every time. A few vocal moments also benefited from a similar approach — some of Freya Pocket's 'Maxed Out' vocals, for example, and the first 'Pow!' from Pete Pocket in the run-up to the third rapped verse.

Quickly fading in some of the other parts against this rough balance, though, swiftly demonstrated that the snare would need a helping hand if it was going to compete with the general sonic mayhem of the production as a whole — although it had lots of attack, there wasn't much in the way of aggression or sustain. Dialling in some tube distortion and limiting, courtesy of Silverspike's Ruby Tube, immediately gave the sound more attitude and hold, but this was at the expense of some attack, so I compensated for that with Stillwell Audio's Transient Monster (which works in a similar way to SPL's well-known Transient Designer). The same company's Vibe EQ (which is my current favourite 'character' equaliser) was used to mould the sound further, with a boost of around 5dB at 820Hz for body, and another 6dB of high-shelving boost above 8kHz to sharpen the edge. Finally, a little Aphex-style high-end psychoacoustic enhancement topped things off, bringing the sound further forward than any EQ tends to be able to do on its own.

The heavy processing I was now using on the snare's close mic inevitably also had an effect on the other instruments spilling into that mic, but luckily these side-effects, for the most part, turned out to be beneficial to the overall drum sound. However, the hi-hat was now being pulled much higher in the mix by the action of Ruby Tube's limiter, so I turned down the hi-hat's close mic, to keep it from eating the rest of the kit for breakfast!

At this point the drums were holding together as an instrument, but still had a rather brittle, digital feel, and I felt that something crunchier and more underground was required to fit the hip-hop frame of reference. So, bussing all the drums to a single track in Reaper, I applied some parallel compression (using Buzzroom's freeware Grancomp plug-in, which I merrily hammered to 12dB gain reduction at a 2:1 ratio, but only mixed in at a fairly low level) and then smacked them hard into the end-stops of Jeroen Breebaart's Ferox tape-emulation plug-in. This really fattened things up, and rounded the hat and cymbal transients in a very nice way. I had to tweak the snare processing after adding the tape saturation, though, as it was losing a touch too much snap.

However, there were plenty of other concerns to tackle. The rhythmic element of the sub-bass part was not at all in time with the drums. Chickening out of the kind of note-by-note editing that I'd have needed to do to pull the notes into line, I low-pass filtered at 100Hz instead, to reduce the audibility of the few higher-frequency transients, and then ducked the sub-bass level by a couple of decibels between the kick-drum hits, to make the bass rhythm appear to follow the drums more closely. The ducking was implemented using Reaper's Reagate plug-in, with its side-chain input fed from the kick-drum track. Although Reagate doesn't have the traditional 'range' control, it does allow you to mix in the ungated signal with the gated signal to similarly soften the gating action.

The filter sweeps of the other synth were causing mix-balance problems, on account of the resultant dramatic shifts in harmonic content. In these situations I usually compress the part heavily, often across multiple frequency-bands, and in light of the already powerful drum parts here I pushed this concept as far as I ever have, crushing the channel with about 20dB of limiting from Buzzroom's multi-band BuzMaxi3 limiter.

This did a good job both of solidly anchoring the level and of flattening out the worst frequency-domain peaks, but there was still a long way to go. As it stood, the drums were psychoacoustically masking all the synth's higher frequencies and making it sound tame and muffled, so I decided to dramatically boost those to get them to cut through more clearly. Vibe EQ again did the honours with 4dB cuts at 110Hz with the low shelf, and at 820Hz with the mid peak, as well as 7dB of high-shelf boost at 2.5kHz upwards. This certainly helped to bring the sound forward, but I still wanted more bite, so added some extra distortion from another instance of Ruby Tube and a brutal 9dB peaking boost at 3.2kHz from Stillwell Audio's emulated Neve EQ plug-in, 1973. This amounted to a lot of heavy processing — and much more than I'd recommend using most of the time — but I felt that it was justified in the context of this particular mix.

Listening to the two bass guitar parts, the cab track seemed to have a much more appealing character to it, so I didn't even use the DI in the end. I often tend to use DI tracks to provide a more predictable low end, but that was already coming from the sub-bass synth anyway. Primarily it was the guitar/amp combination's mid-range that I thought was most likely to carry through this mix, so I cut away everything below 350Hz using ReaEQ and boosted a 16dB peak at 1.1kHz, using the more characterful Antress Modern Booster EQ (apparently modelled on early SSL console EQ, which adds a nice kind of grit with big boosts). This whittled-down sound slotted into the mix pretty well, although it stepped on the toes of the filtered synth until I dipped 7dB out of the guitar in the disputed area (around 3kHz).

The SM58, like most dynamic mics, isn't as detailed at the extreme high end as a typical studio vocal condenser, so my next stop was dialling in some 'air' with a few decibels of high-shelving boost at 17kHz, although both the slope of the shelf response and the amount of boost varied between the two parts. The next thing I noticed was that the mid-range wasn't yet coming through well enough in the context of the mix, leaving the vocals sounding a bit soft — which wasn't a good thing.

A dose of peaking boost at around 1kHz helped a lot with that. Initially I just used the URS's '1951' Pultec EQ emulation, but once I was getting into the right ballpark I began flipping the different bands to other models, to see what might work best in each case. As I'd expected, at the high end the Pultec was my preferred choice (it's great at brightening a sound without showing any strain), but the mid-range boost certainly benefited from a change on both vocals, to the markedly less polite '1970' (Neve 1073) and '1980' (SSL E-series) options respectively. Finally, I upped the transformer-input stage model's character setting to make the sound a bit larger than life — I can never seem to get enough of transformers! This is one of the things I really like about Console Strip Pro: it sounds good however much you use it, but gives lots of subtle tone-shaping options to play with on those occasions where you have the time to harvest a few more of the diminishing returns.

De-essing was an essential because of the extra high-end (Digital Fishphones' Spitfish to the rescue, as usual), following which I compressed both vocals heavily, aiming to maximise the audibility of the lyrics. The URS plug-in's Urei 1176 model in its 'All Buttons' mode was an important part of the recipe (that one's always a stalwart when it comes to hurricane-force vocals), but I also followed that up with George Yohng's W1 limiter (apparently a Waves L1-inspired algorithm), to pin the levels fairly comprehensively throughout. Nonetheless, the vocals required plenty of fader automation in the final stages of the mix, because some of the vocal syllables weren't delivered with quite the same attack as others, and little level boosts were necessary to maintain the rhythmic momentum.

For the numerous rapped backing-vocal parts, mostly it was a case of high-pass filtering across the board and some judicious dynamics processing from W1 or Dirtsqueeze, although I also used Ruby Tube or MDA's Combo plug-in on some tracks to make them slightly 'rowdier'. Processing the female vocal parts followed similar lines, although the high-pass filter was set higher, at about 670Hz, and I relied on MDA's simple Limiter plug-in for most of the dynamics processing.

The decks parts were interesting to work with, because they don't have as much prominence in most music as they have here. Scratch DJ'ing actually has quite a lot of dynamic range, and I found the tracks all needed a good deal of evening out before I could find a suitable fader level for each one. For most of them I ended up trying out the new Rocket compressor from Stillwell Audio at 20:1, dialling in its most ridiculously fast attack time, and setting its release just shy of the point of distortion, to really tweak up the smaller details in the tracks.

The turntable tracks included the sampled phrase that forms the backbone of the song's middle section, and because this was more pivotal to the sound as a whole I spent a little more time dealing with that. The main problem was that it was in mono, and it seemed to me that it wanted to feel a bit wider in the mix, so I pulled up an instance of Silverspike's Room Machine 844 to try to give it a bit more space. This specialised reverb simulates the effect of playing your sound in a room and then recording it through a stereo pair of mics, and tends to work really well when you need some feeling of real space around an otherwise dead-sounding sample or close-mic, but without any obvious reverb tail.

As usual, it came up trumps here, and really helped pull the sample together with the drum sound. I kept the room-size and miking-distance settings fairly low, and balanced the amount of damping in the simulated room using the 'wet' effect-level control. This, and some hefty gain-reduction and saturation from Reaper's built-in Dirtsqueeze plug-in, helped make things nice and solid-sounding, while a little of Reaper's Exciter pushed the sound forward in the mix without my having to fade it up too far.

There remained only a bit of automation work to guide the listener's ear towards the most important parts of the mix: the drum fills, exposed bass moments, little backing vocal phrases, and so on. In some mixes this isn't really necessary, but where everything's fighting to be right in the listener's face, you need a little 'crowd control' to prevent things getting confused. It's important to remember that you can only start doing this kind of work once you can hear the results in the context of any gain changes being created by the master bus compressor: if you tackle this automation before you compress the master bus, you're likely to have to redo a lot more of it once the compressor is in line.

Referencing the mix against some tracks from NERD and Jay Z showed that the mix needed more extreme high-end, and a little less of the sibilance region. I tackled that by switching in the EQ on my master-bus instance of Console Strip Pro. A low-resonance 5dB high-shelving boost at 20kHz, and a 2.5dB cut at 6.6kHz, balanced the overall tonality more appropriately. I could, of course, have EQ'd the individual tracks to achieve these changes, but why do this with lots of lower-quality EQ plug-ins when you can make a better-sounding job of it with a single high-quality processor on the main output?

The final stage was to push the track to an appropriate loudness level, which I did by combining a few decibels of straight clipping with four decibels of limiting from BuzMaxi3. Using clipping to take some of the strain off the limiter is very common in commercial releases in this kind of style, as it tends to keep a punchier sound, albeit at the expense of slight timbral side-effects on the drums.