The Southern Wild: our engineer reminds us that you don’t have to cram every mic signal into your mix just because it’s available.
There’s something special about recording a group of musicians all together in one room. You might not capture the most technically perfect performance from everyone, but any deficiency in the parts is made up for by the excitement and naturalness of the whole. It’s by far the quickest and most engaging way to capture what a band is all about. At the same time, though, recording ‘live in the room’ introduces technical challenges, especially when the room in question belongs to a house rather than an acoustically treated studio. These are compounded when the band are self-recording: what might be obvious to an engineer listening in an isolated control room is easy to overlook when you’re focusing on your own playing and singing. So it was that Americana three-piece the Southern Wild had emerged from an enjoyable recording (and filming) session with a multitrack that was proving frustratingly hard to mix, even though all the individual elements were cleanly recorded.
When you’re not in a position to monitor objectively, or to spend hours working on mic placement, there’s a temptation to put up every mic you can lay your hands on in the hope of scoring at least some hits. Listening to the multitrack for ‘Curtis Stirling’ suggested that something of the sort had gone on in the Southern Wild’s sessions. Although the song featured only four instruments and one vocal, there was a total of 12 audio tracks, with both bass and guitar represented by DI and two miked tracks, and cajon miked both inside and out.
Having recorded lots of mics, there’s a natural tendency to want to use them all, and I wonder whether that might have contributed to some of the difficulties singer Drew Stephenson was having with his own mix. He’d got a pretty good basic balance of instruments, but there was a general muddiness to the sound and more than a whiff of comb-filtered, roomy spill, suggesting that too many ingredients had been thrown into the pot.
Don’t get me wrong: there are times when double-miking a source can be very worthwhile. Most obviously, of course, it’s a basic requirement if you want that source to have any stereo width. Some engineers also do it in order to exploit the tonal properties of different microphones, or to capture important but separate elements of the sound produced by large instruments such as the double bass. However, a lot of people put up multiple mics with the rather more vague goal of “having options available at the mix”. This is a mixed blessing to say the least, and all too often it produces two or three unsatisfactory sounds rather than one really good sound.
I’m not sure how the Southern Wild arrived at their mic choices and placement but, in the end, I don’t think they gained much by leaving these options open. Neither of the two mics on the bass amp had managed to obscure its fundamentally honky and cheap sound; and the two mics on the cajon had been placed so close that neither really sounded much like a cajon. Drew’s acoustic guitar had been miked in stereo, which was a reasonable idea, but both mics had also picked up a great deal of bleed from his vocal and the other instruments.
When you’re faced with a combination of room and close mics, I find it’s quite often the case that each individual source needs to favour one or the other. That is, for any given instrument, the close mic can supply the main sound with support from room mics and spill, or the close mic can be used to subtly assist that instrument’s presentation in the room mics. Problems tend to arise more frequently if an instrument’s presentation in the mix comes equally from two or more different mics.
With this in mind, I began the mix by trying to identify one track in each case that would provide the main source for each instrument. Listening first to the room mics (see 'Careful With The Sides' box), it was clear that the band’s makeshift acoustic treatment had made their recording space sound pretty decent, at least down to 300Hz or so. However, the room mics hadn’t picked up the sort of balance of instruments you’d want in the final mix: the cajon was much more prominent than the other sources. It also sounded much better in the room mics than on either of its close mics — some instruments just need a bit of space for their sound to bloom and come together.
The balance in the room mics wasn’t quite so skewed to the cajon that I could get away without using its close mics at all, but this was definitely a case where they would be best off operating in a supporting role. However, figuring out the best way to make this work wasn’t as straightforward as I’d hoped. Somehow, the close miking had produced two tracks which sounded more like abstract electronic beats than a plywood box being struck with fingers and palms. Both mics also emphasised a very strong resonance around 120Hz.
I managed to tame the resonance by time-delaying the cajon’s inside mic by about 3.5ms, so that a helpful phase cancellation occurred when it was combined with the outside mic. This improved the sound of the cajon mics in isolation, but I wasn’t going to be using them in isolation. The cajon sound in the room mics had plenty of snap and buzz; what it lacked was mid-range punch, so I added some fairly heavy processing to the close mic bus to focus the sound on this missing region. One compressor and two distortion plug-ins later, I had an overall cajon sound that both sounded like a cajon, and had some substance and power to it.
The cajon’s prominence in the room mics made clear that for all the other instruments, it would need to be close mics or DIs that took the lead role. But which ones?
It didn’t take much listening to convince me that that bass sound would have to come entirely from the DI, as the amped sound was plain bad. In all honesty, the DI’d sound wasn’t amazing either, but it was at least clean, spill-free and generally amenable to being improved. It took quite a bit of compression to bring the dynamics into line, and I found that instead of a conventional amp simulator or distortion effect, FabFilter’s multiband Saturn saturation plug-in did a better job of flattening out the sound and removing the remaining honkiness.
With a heavier heart, I also decided that the DI would have to carry the bulk of the acoustic guitar sound, as there was simply too much vocal spill on the guitar mics. Pickups in acoustic guitars can be pretty hard work at the mixing stage, but this one sounded reasonably natural, at least once I’d scythed away the usual excess of hard-sounding mid range. Being a sentimental sort, I kept in one of the guitar mics at a relatively low level; I’m not sure its contribution was really noticeable in the end, but I convinced myself that it made the sound a little more natural.
There was no choice on offer when it came to Drew’s vocal, but there was a need to do something about the way it sounded. He’d sung into a Shure Beta 87 stage mic which, I suspect, had been placed a bit too far away, because the basic tone was thin and harsh. This is a problem I’ve had to address in several previous Mix Rescues, and I’ve found it often doesn’t respond well to static EQ, because not every word or phrase is equally affected. A multiband compressor is usually more effective, and in this case, it took five bands in FabFilter Pro-MB to bring Drew’s vocal into line. All but one were set flat, but with hair-trigger threshold values so they’d duck various bits of the mid range at the first hint of harshness. The lowest band was also configured as a compressor, but when no compression was taking place, it applied a generous boost centred around 400Hz to thicken up the basic sound. On top of that, I found I needed to automate a 250Hz EQ band to top the low mids up in some sections; and though I’d used a conventional compressor as well, it was no surprise to find that both the vocal and instruments needed some fader automation to maintain the right balance through the song.
The remaining instrument was a shaker overdub, which Drew had found hard to place in his mix. Over the years I have come to be increasingly brutal in processing this sort of percussion, which has a tendency to sound spiky and thin. In this case, I led the assault with a huge EQ boost in the mid range and a fairly heavy degree of saturation. It soon waved the white flag and sat meekly in the track.
I’ve never managed to train myself to ‘mix into the compressor’ as some engineers like to, but at some point in the process, I’ll usually put one across the stereo bus as an experiment. Sometimes it stays, sometimes it goes, rarely does it apply more than a decibel or two of gain reduction. On ‘Curtis Stirling’, it stayed.
In my experience, mix-bus processing is more useful for shaping the tone of the mix than for dynamic control, the general aim being to fill out every last corner of the mid range. Here, I used SoundToys’ Radiator and Slate’s VTM to add some general warmth, and pushed the 500Hz-3kHz range into a couple of bands of multiband compression to squeeze as much as possible out of it before the mix became tinny or honky.
Thankfully, the Southern Wild’s feedback on my first attempt was not of the “Can you cut half a dB at 267Hz on the second triangle overdub?” variety. Drew’s main request was for less reverb on his vocal, so as not to undermine the feeling that the listener is hearing a band playing in a living room. When the focus in Mix Rescue so often is to overcome the compromises of home recording, it’s nice to embrace the reality for a change! The decision to record live was exactly the right one for them, and even if it introduced a few challenges at the mix, ‘Curtis Stirling’ has a directness and immediacy that no overdubbed studio multitrack ever could.
The audio files on this page accompany the Mix Rescue feature in SOS July 2016, in which I reworked the track ‘Curtis Stirling’ by the Southern Wild.
The Southern Wild had recorded ‘Curtis Stirling’ live in a domestic room, and the aim was to improve the general mix without losing that live, raw quality.
The room mics had been recorded as a Mid-Sides pair with non-matching mics. As you can hear, the Cajon was prominent in the room sound.
As you can hear in the first of these examples, of the three different presentations on offer for the bass guitar, the DI was the most promising. The second example is the DI track as heard in the mix.
Two mics had been placed on the cajon. In these examples you can hear them each in turn, then together. Note how in the unprocessed example, the combination of the two mics reinforces the resonance at 120Hz.
Drew Stephenson’s raw vocal sounds harsh and thin; in the remix, I reshaped its tone using a multiband compressor.
Drew Stephenson writes: “As this was the first time I’d tried to record and mix from multiple instruments I wasn’t really sure how to get started. I knew I wanted to try and capture the feel of the three of us playing together in the room, so I tried to focus on using the room mics as much as possible and then bring up the close mics as needed. One of the challenges was that, having set up multiple mics to give myself plenty of tone options, I then tried to use all of them...
“Getting the tone balance right between the bass and cajon was particularly difficult as these aren’t instruments I’m as familiar with.
“Obviously we’re much happier with the remix; the sound is fuller without being congested, the bass and cajon occupy their natural frequency areas much better and the dynamic range is much better. I also like the fact that the vocal sounds a lot more like me! And I now have a good template to work with for the other songs from the session.”
One of the classic mixing conundrums is that some band line-ups just don’t lend themselves to stereo panning. The convention that drums, bass and lead vocals should be panned centrally is so ingrained that any departure from it sounds like a perverse attempt to return to the late ’60s. But if you’ve only got one other main instrument, that leaves you with a dilemma. Do you pan the extra instrument to one side, keep the core tracks in the middle and accept that your mix will be unbalanced? Do you put everything straight down the middle and mix more or less in mono? Or do you try to compromise the Holy Trinity of centrally panned sources in order to have things out wide on both sides?
This was a dilemma that definitely presented itself in the Southern Wild’s recording. Drew Stephenson’s vocal was the most important thing in the mix, and obviously needed to go straight down the middle. But which of the instruments could be panned away from the centre? The bass? The cahon, which was the main rhythmic element? Or the guitar, which, conceptually at least, belonged in the same position as Drew’s voice, since he was playing it?
The purist approach would be to maintain the panorama that the room mics had captured. In this case, though, they weren’t much help, as they hadn’t really captured one. Even when I temporarily ramped up the Sides signal to exaggerate the stereo width, it didn’t feel that any of the instruments naturally belonged on one side or the other.
In the end, I decided that the most natural approach was to hard-pan the guitar one side and the cahon’s close mics the other way. This didn’t in any way match what was going on in the room mics, but in timbral terms, it meant that there was a left-right balance between the two instruments that were carrying the song’s mid range; and unlike a full drum kit, the overall ‘smaller’ sound of the cahon didn’t seem odd when pushed to one side.
One of the challenges Drew Stephenson had faced in setting up the Southern Wild’s recording session was that he didn’t have a good matched pair of microphones. Very sensibly, therefore, he had chosen to set up his room mics using the one stereo technique that doesn’t require matched mics, namely Mid-Sides. Acknowledging the importance of the room mics in a live recording, he’d reserved his two ‘best’ mics for this role: an SE Electronics SE2200 capacitor mic for the Mid and a custom Xaudia ribbon for the figure-of-eight Sides.
This was a good decision, but it meant that there was quite a contrast between the tonality of the two, with the bass and low mids much more strongly represented on the Sides track than on the Mid track. And this, in turn, meant there was a lot of muddy, out-of-phase low end on the decoded left/right stereo room track, which is something to be wary of at mixdown.
It’s very hard to predict how such a ‘wide’ bottom end will behave in any given listening environment, and it will simply disappear when the track is auditioned in mono, so in my remix, I high-pass filtered the Sides track fairly savagely. To try to match the tonality of the two mics a bit more closely, I also boosted the Mid’s mid and the Sides’ high frequencies.