Mix Rescue: War Kites

The War Kites guitar recording station. Somewhere in that pedalboard, I suspect, lurks a source of unwanted noise...

War Kites: We help a talented indie band achieve a more polished sound from their home recordings.

Time was when entire albums would be tracked and mixed in one stretch of studio time, by the same engineer, with the artist and producer in permanent attendance. Music production today is much more of a piecemeal affair. Laptops are carted from studio to studio, files are piped over the Internet, and mixing and mastering are often carried out remotely in unattended sessions. This has its advantages, but there are down sides too. When music production becomes a distributed process, with all communication handled electronically, the potential for misunderstanding and confusion grows exponentially — which is why this month’s Mix Rescue took a little longer than usual!

It began with a post on the SOS Forum, with War Kites guitarist Sam Morris seeking reassurance that the band’s rough mix was heading in the right direction. Most of us who listened to the track felt that there were some issues that needed addressing. In particular, there was general agreement that in the chorus sections, when the distorted guitars kicked in, the mix was folding up and getting smaller where it really needed to do just the opposite. I suggested that they might like to send it over as a Mix Rescue candidate, and soon enough, I was downloading the best part of 70 audio files.

The size of the project was a surprise, as ‘Stop The Flow’ was not a particularly complex arrangement, consisting of drums, bass and half a dozen guitar parts. There were, it turned out, two reasons for the file bloat. One was that nearly all of the guitar parts had been double-tracked, and some had been stacked as many as five times; and in every case, two mics had been recorded. There was only one bass part, but that had been tracked through four different mics and a DI! The other reason was that the drum tracks had been exported from XLN Audio’s Addictive Drums 2 virtual instrument, and thus included a comprehensive selection of close, overhead and room-mic tracks.

Booty Guitars

A small selection of War Kites guitar tracks in the Pro Tools Edit window. Note how all of them are closely trimmed, with any non-playing sections muted, to minimise noise. The track at the bottom of the screen has been switched to Warp mode, making visible my timing corrections.Turning first to the guitars, it quickly became clear why the band’s own mix had run into trouble in the choruses. All the guitars had been tracked with both a Shure SM57 and an Audio-Technica AT4033 capacitor mic, but a combination of amp settings and mic placement meant that every single guitar track was plagued with much too much bottom end. To add to the fun, nearly all the guitar parts also came with a generous side helping of noise; even on the ‘clean’ guitar tracks, this was so prominent that I couldn’t even find a gate setting that would reliably separate the wanted notes from the unwanted buzzes and hums. A third factor that I thought might prove an issue was that many of the clean guitars had been tracked with quite a lot of amp reverb: not necessarily a bad move, but certainly one which would close off some options at the mix.

My heart sank at the thought of running 20-odd guitar parts separately through a noise-reduction processor. The noise varied enough that different settings would have been required on each track, and the whole business would have taken hours. Since I was having to cut huge amounts of low end from all the guitar parts in any case, I also added low-pass bands variously set between about 4 and 6 kHz, then manually trimmed all the regions within Pro Tools, cutting out the lengthy rests in every guitar track and muting them. This tamed the worst of the noise, and in the context of the mix, it wasn’t audible.All of the guitar tracks had a lot of unwanted low end, and most were also very noisy. This is a representative sample of the EQ settings it took to bring the clean guitars into line!

Double Trouble

Many self-recording artists employ a lot of double-tracking, perhaps seeing it as a way of compensating for insecurities about their playing or engineering. My own feeling is that when the performance and sound are on the money, doubling usually adds little of value; and when they’re not, it often serves to multiply any problems. With this in mind, I was poised to chuck out all of War Kites’ guitar doubles, especially as they mostly used exactly the same guitar sounds. In the event, however, some of them did turn out to be useful.

When you have a lot of tracks to deal with, organisation is critical, so I began by allocating each set of guitar tracks to its own Mix and Edit Group in Pro Tools. I then created a bunch of stereo aux tracks and routed each of these groups to a different one, allowing me to process, mute or automate each set of guitar parts using a single mixer channel.

The main musical element in the verse sections of the song was a clean, descending guitar riff that had been recorded five times through both guitar mics, giving me potentially 10 nearly identical tracks to play with. The capacitor-mic tracks seemed to have a bit more substance to them than those tracked with the SM57, so I chose one of them more or less at random: on its own, though, it sounded a bit naked and small next to the epic drums, so I experimented with various ways to make it sound bigger.

Introducing one of the doubled takes helped, with its two mics panned hard left and right, but only up to a point. When I brought these mics up too far, or introduced the other sets of doubles, everything became a bit vague and the notes lost their definition, especially when the mix was heard in mono. So, to thicken the sound further, I duplicated the original ‘lead’ track and slung it through a slightly edited version of the ‘Fuzz Mod’ preset in SoundToys’ Primal Tap.This patch from SoundToys’ Primal Tap proved a more rewarding way of thickening the clean guitars than using all the band’s double-tracks. This, as the name suggests, adds some chewy, heavily bandwidth-limited distortion fed through very short modulated delays, the upshot of which is a meatier, middlier, wider sound with some subtle motion.

Finally, with quite a bit of amp reverb already baked into the sound, it would have been redundant to add more, but stereo delays added to the richness of the brew without pushing the guitar into the background. I ended up with three of these on different aux sends: a fairly simple eighth-note delay in SoundToys’ Echo Boy, and a couple of more complex patches from Eventide’s H949 Dual Harmonizer and H3000 Band Delays. All were used sparingly on different guitar parts in different sections of the song.

Meanwhile, In The Chorus

Similar guitar sounds had been used for most of the clean parts, but there were a couple of exceptions. A driving rhythm part in the chorus required little processing beyond some robust mud removal, whereas the moodier, semi-half-time section of the song featured a nice ambient sound aptly labelled ‘violin guitar’ which cried out for some unnatural treatment. Again, Primal Tap came up with the goods, courtesy of a preset named ‘Bounce Slap Shimmer’.

When it came to the fuzz guitar that was the main feature of the song’s chorus, the big problem was definition. There were two takes of this riff, both using the same sound; as well as being overwhelmingly bassy, both were so heavily distorted as to degenerate almost into white noise at times. Below 400Hz or so, the bottom end was overblown to the point where it simply contained nothing that was recognisably musical; but removing it altogether left a gutless guitar sound that was inconsistent from note to note. It took fairly radical equalisation, and a certain amount of multi-band compression, to arrive at a sound that was mixable. Somewhat by accident, I also discovered that my ‘Fuzz Mod’ Primal Tap setting helped to thicken up the missing mid-range, which was lucky.

Race To The Bottom End

Removing all the low-mid mud from the guitars meant that there was now space for the bass — all five tracks of it! I decided there probably was no need to use all four mic tracks, so muted three of them and kept the one which seemed to me to have the most solid low end. With the rest of the track shaping up to be quite rocky and aggressive, however, the amp tone was coming across as rather polite, so I set out to add some muscle to it by shaping the DI track. Usually I like the SansAmp PSA-1 plug-in you get free with Pro Tools for this job, but in this case, I ended up with two different settings from Scuffham Amps’ excellent S-Gear (which is designed as a guitar amp simulator, but can work well on bass). I used automation to fade these up or down depending on how much else was going on at any given point in the mix.

Distant Drums

I’ve received quite a few potential Mix Rescue projects where bands have tried to recreate natural-sounding performances with sampled drums, and most of them have been pretty much un-rescueable. The drums are so fundamental to any rock or pop arrangement that if the programming or sounds aren’t up to the mark, there’s nothing much you can do about it at the mix. So full marks to War Kites drummer Connor Di Leo for coming up with one of the most convincing fake drum parts I’ve ever heard, in a musical genre where this is particularly difficult to get right. It’s no surprise to discover that the secret here is hard work: “I wanted them to sound more professional than out last album,” says Connor, “so I wrote the parts, made sure I could actually play them (by rehearsing them) and manually, note per note, input them as MIDI notes into the track.”

Unsurprisingly, there wasn’t much wrong with the raw output from Addictive Drums. However, when asked for mix references, the band had cited modern rock acts like Queens Of The Stone Age and Foals, and in comparison to some of the drum sounds on offer from those bands, the sampled kit they’d chosen seemed to me to lack a little bit of thump and weight in the low mids. This instance of FabFilter’s Pro-MB was set up as an expander, to apply a momentary boost between 100 and 400 Hz every time the snare was hit.

I’ve lately become a fan of IK Multimedia’s Black 76 and White 2A plug-ins, which form part of that company’s T-Racks Custom Shop mastering package. I’m not sure exactly how faithful they are to the 1176 and LA2A on which they’re modelled, but they certainly have the ‘hair’ that is sometimes lacking in plug-in compressors. In this case, the Black 76 seemed a good choice for adding grit and ‘smack’ to the kick and snare close-mic channels. To restore the missing weight in the low-mids, meanwhile, I used an instance of FabFilter’s invaluable Pro-MB multi-band dynamics processor across the snare track, with just a single band covering the 100-400 Hz range. This was configured as an expander, with the fastest possible attack and release times, so that it added a momentary boost of 4dB or so in this range every time the snare was hit, the idea being to add ‘thump’ to the attack of the snare without muddying its decay.

On the drum bus, meanwhile, I employed the opposite trick a bit higher up the frequency spectrum: a Pro-MB band centred around 500Hz but configured as a compressor, so that it was boosting all but the loud peaks. The same company’s Saturn multi-band distortion added further low-end thickness, while Slate’s FG-Red compressor and Virtual Tape Machines helped to make the tracks hit harder and sound a little less pristine. The main ingredients of an epic snare reverb: FabFilter’s Timeless delay and Overloud’s Rematrix convolution reverb.

Thirsty Snares

The raw sound from Addictive Drums 2 was naturally pretty dry. This was in keeping with the style of some of the band’s references, especially Queens Of The Stone Age, but it was soon clear that retaining that dryness at the mix wasn’t going to work. For one thing, even the most skilfully programmed drums tend to sound inhumanly consistent, and the more dry and upfront the snare sound in this case, the more noticeable were its sampled origins. For another, the clean guitars had mostly been tracked with dollops of amp reverb, so keeping the kit sound dry led to an odd front-to-back imbalance where the lead instruments sat behind the rhythm section. Finally, I was hoping that some additional reverb would help add further meat to a snare drum that wasn’t yet in Dave Grohl territory.

When you have the resources of a modern DAW available, I think it makes sense to have separate reverbs for the drums, because these can then be returned to the drum bus and fed through any processing you’re applying to the entire drum mix, helping them to feel more like they belong to the drum sound. I spent quite a long time tinkering with the drum reverbs over the course of the mix, and ended up with no fewer than four stereo aux channels devoted solely to snare and tom reverb. Two were impulse responses from HOFA’s IQ-Reverb, both of which could be categorised as ‘ambiences’ with very short decay times, helping to thicken out the sound without making it conspicuously reverberant. (I didn’t really mean to use two, but having set them both up, I couldn’t decide which I liked the best.) Another was a fairly standard medium plate from Exponential Audio’s R2.

In combination, these three snare reverbs got the basic drum sound to a point where it had plenty of power, yet stayed far enough back in the mix to make the guitars feel like the lead instrument. However, I felt as though the drum sound should also change to mirror the arrangement, where the hard-hitting chorus alternated with other sections that were much more floaty and dreamlike. Connor’s drum programming cleverly reflected this, hinting at switches between straight 4/4 and half time without ever fully committing to one or the other; and in the dreamy middle section, his busy snare work gave way to a simpler pattern driven by a single prominent snare strike on the offbeat. I wanted to explore the idea of enhancing the dreamlike feel by applying an over-the-top reverb just to these snare hits. Overloud’s Rematrix convolution reverb has a good selection of larger-than-life drum patches, so I dialled up one called ‘Huge Snare’; this didn’t exactly do the trick on its own, but it occurred to me that by placing a compressor and a delay plug-in in series with it, I could turn it into something altogether more ambient and washy.More low-mid weight was added to the drum mix by this setting from FabFilter’s Saturn distortion plug-in.

I liked this ‘Epic Snare Verb’ effect so much that I actually applied it all the way through the song, albeit with extensive automation to reflect the fact that there was less space for it in the choruses. However, I only really wanted it to emphasis the main snare strikes, so rather than feeding it from the main snare track, I created a duplicate and used Pro Tools’ Strip Silence function to remove the ghost notes and other low-level snare work.

Narrowing Down

As my mix took shape, I began to feel that the stereo overheads presented a disconcertingly wide picture of the kit as a whole. I experimented with panning them less than 100 percent left and right, but found I preferred the sound that I got by simply muting the left-hand overhead mic and treating the right overhead as a central mono drum mic. Leaving the room mics stereo then gave some ‘size’ to the kit without undermining the impression that it was located solidly in the middle of the mix.

The final touch to the drum sound came about as a result of the band’s feedback on my initial mix. They suggested that some sort of effect could be applied to the low-key drum part in the first eight bars, so as to enhance the ambient feel of the song’s introduction and ensure a sense of impact when the main drums arrived. This was clearly an excellent idea, but making it work proved harder than I’d expected. Applying yet more reverb just turned everything to indistinct mush, while EQ and filter plug-ins either seemed to make the drums disappear altogether or stick out far too much. My eventual solution involved using Avid’s trusty Lo-Fi plug-in, routed through Eventide’s Instant Phaser and SoundToys’ Pan Man auto-panner. With anti-aliasing turned off in Lo-Fi and the sample rate parameter automated to move slowly, the results sounded something like a cross between a crunchy saturator and a resonant filter, adding movement and contrast without taking over the mix.

Wait, You Mean...

I mentioned at the start of this article that the mix process had suffered from a communication breakdown. This only became apparent once I’d sent a draft mix off to the band. Their reply? “We do have some feedback, but we thought you might like to try and mix in the vocals (which we recorded today) before we make our final comments.” In other words... it wasn’t an instrumental after all!

I still don’t quite know how we contrived such an epic misunderstanding, but 24 hours later, I was downloading separate lead vocal tracks for verses and choruses, plus some chorus backing vocals. A sensibly placed capacitor mic had been used to capture clean close takes, which all sounded fine, but the band had also put up an SM57 as a lo-fi room mic. This was too noisy to use high in the mix, but did add some nice character when mixed in low.

Luckily, as it turned out, adding the vocals didn’t require huge changes to the rest of the mix. With ‘Stop The Flow’ being a full-on alternative rock track, it wouldn’t have been appopriate to sacrifice power in the other instruments to make way for a dominating lead vocal. Instead, mixing it became a matter of ’tucking in’ the vocal to the backing track, both dynamically and tonally. This was achieved using a combination of level automation, reasonably aggressive compression, an EQ boost at around 3kHz, and a separate parallel effect track where I added mid-range crunch using the aforementioned SansAmp PSA-1 plug-in. This too was automated, changing the setting of the Crunch parameter to cut through more in the dense sections of the mix. The lead vocal got its own reverb — a large plate from HOFA’s IQ-Reverb — as well as some spot delay effects that came in and out for different sections of the song. Of all the plug-ins that come with Pro Tools, the humble SansAmp PSA-1 has to be the most versatile. Here, it’s adding crunch to the lead vocal.

It was all sounding good, but when I sent my first vocal mix to the band, I ran into another of those awkward ‘I wish I’d asked’ moments: I had actually managed to drop the vocal parts into the wrong place in the track, so that they were a bar ahead of where they should be! In my defence, they actually made musical sense in both places, but I should have checked. Once I’d sorted that out, along with a couple of more minor changes, we had a mix that everyone was happy with. (Either that, or the band were simply too exhausted to pick me up on anything else.)

Go With The Flow

Once upon a time, the maximum size of any given mix would have been dictated by some fairly immovable limitations: the available number of tape tracks, mixer channels and outboard processors. In the DAW age, though, it’s all much more open-ended, and the purist in me feels slightly uncomfortable about the way in which my mix of ‘Stop The Flow’ seemed to grow and grow. By the time it was finished, my Pro Tools session had more than 30 aux channels going, with so many plug-ins that it wouldn’t even play back unless I disabled the CPU-hogging output limiter.

Ultimately, though, most of the work that these channels and effects is doing is either utilitarian — applying corrective EQ, tuning vocals, bussing collections of doubles together — or restricted to a single musical part or song section. Some of them could be muted without making the mix fall to bits, but they’re all doing something positive, and the results would be poorer without them. Truly, the devil is in the detail, and once the basic hurdles of balance and tonality have been cleared, it’s the accumulation of many details that makes a track sound ‘produced’. Or, if you get it wrong, over-produced...