Rescued this month: based in Dunedin, New Zealand, the Broken Heartbreakers consist of John Guy Howell (vocal, guitars, piano and vibes); Rachel Bailey (vocal and guitars); Jeff Harford (drums) and Richard Pickard (electric and double bass).
This month’s band threw down the gauntlet for our engineer — could he improve what was already a decent mix?
In a series titled Mix Rescue, it’s no surprise that the focus is usually on solving problems. Whether it means unravelling misplaced plug-in chains or compensating for mistakes made at the tracking stage, fixing it in the mix is something we all have to do from time to time. There is definitely satisfaction to be had from salvaging a flawed recording, but in an ideal world, it’s not what mixing should be about. When you have great material that is well produced, brilliantly performed and thoughtfully recorded, the job of the mixing engineer is not to overcome its weaknesses, but to build on its strengths.
Most people who pick up the Mix Rescue hotline do so out of desperation, having found themselves in a hole they can’t see a way out of. So I was intrigued when New Zealand-based engineer Mike Stoodley emailed with a completely different proposition. His living-room sessions with Dunedin-based band the Broken Heartbreakers had yielded finished recordings that both he and the band were extremely happy with. But, said Mike, “as a long time Inside Track and Mix Rescue follower, I still have an itch to scratch: what would the SOS guys have done with it? What would their tweaks be that would make this even better?” We were, in other words, being challenged to improve upon Mike’s own mix — a mix with which the band had been intimately involved, and which they clearly loved to bits. No pressure...
No Low-hanging Fruit
When I first listened to Mike’s mix, I was secretly hoping to spot obvious flaws which would make for an easy win on my part. No such luck. His confidence was entirely justified. This was a very good mix of a very nice recording, where nothing jumped out as sounding wrong, and plenty of things jumped out as sounding exactly right.
Rachel’s gorgeous alto singing is the foundation on which the Broken Heartbreakers’ country-folk sound is built, and Mike had positioned the lead vocal exactly where it needed to be: front and centre, without any unnecessary effects or processing to gussy it up. Brushed drum kit, bass and acoustic guitars made for a solid rhythm section, and by blending electric 12-string and vibraphone, the band had created some ear-catching lead parts as counterpoint to the vocal. At first listen, only one thing suggested an alternative approach, and that was the sound of the drum kit: the drums were quite prominent in Mike’s mix, yet most of their energy seemed to be coming through in the upper mids, and I wasn’t getting much sense of weight or ‘thump’ from the kick and snare.
The multitrack arrived as a Reaper project, from which Mike had thoughtfully removed all of his own mix work apart from basic fader settings. I’m afraid I feel about Reaper much as I do about the legal system — I’m glad it exists, but never use it if there’s an alternative — so my first step was to export all the sources as audio files, in order to mix in Pro Tools.
Branching Out
I typically start a mix with the drums, and in this case it definitely seemed the right thing to do, as my early hopes for improving on the original rested with the drum sound. It occurred to me that I might be able to maximise the ‘thump’ of the drum parts by making sure the phase relationships between the mics were solid, so I began by experimenting with various micro-delay settings from Eventide’s Precision Time Align plug-in, auditioning the results in mono on a single speaker. The differences are often subtle, and it’s easy to lose sight of the bigger picture when you start doing this, but I felt that slight delays on the kick, snare and ‘ride side’ overhead mic did bring about an improvement.
Mike Stoodley: “My overall approach was to avoid close miking, so you’ll see that the mics are a good 6-12 inches from the source to capture the full bloom of the source, so to speak. I started off without overheads, hoping that the tom and snare mics would pick up sufficient cymbal and hat, and that the mics would be sufficiently far apart to not have any phase alignment problems. I decided during the tracking that I needed some more cymbal and hat, but couldn’t use normal overheads as the whole band was playing in the same room, so went for a sort of spot-miking, again hoping it wasn’t going to create unmanageable phase problems.”
When I unmuted the second speaker and panned the overheads hard left and right, the stereo image turned out to be a bit wonky, with the kick more prominent on one side and the snare leaning the other way. I preferred the solid feel and stable positioning that I got by keeping the drums almost mono, so ended up panning the overheads just 20 percent either side of centre.
However, the drums still weren’t as ‘thumpy’ as I’d have liked, so I began to turn my attention to the tonality of the individual kit mics. The kick drum’s close mic sounded pretty much like kick drum close mics always do, but the overheads had picked up a strong and fairly nasty resonance at about 350Hz on the kick drum. I didn’t want to simply EQ this out, as it would have made an already light snare sound even thinner. Instead, I used a single dynamic EQ band in FabFilter’s Pro-MB, triggered by a side-chain signal from the kick’s close mic so that it was only ducked when the kick drum was hit.
Conventional EQ’ing of an unappealing kick resonance in the overheads resulted in collateral damage to the overall drum sound, so Sam fed a dynamic EQ band in FabFilter’s Pro-MB with a side-chain signal from the kick close mic, so it dipped the offending frequency only during kick hits.
The kick mic itself had plenty of sub-bass ‘boom’, but was lacking in what you might call the ‘upper bass’ region that’s so important both for warmth and small-speaker translation. I often run kick-drum close mics through saturation or distortion plug-ins for this purpose, and in this case, I found what I was looking for in FabFilter’s Saturn, which allows you to split the input signal into two or more frequency bands and apply different distortion algorithms to each band. With a little bit of work I found a two-band setting that really filled out the anaemic 100-200 Hz zone, while simultaneously downplaying the boxy aspects of the sound a bit further up in the mid range.
FabFilter’s Saturn multiband saturation processor provided a means of adding energy to the kick in the ‘upper bass’ region, improve audibility on smaller playback systems.
I turned next to the snare close mic, which was cleanly recorded, but favoured the ‘tick’ of the brushes hitting the head over the ‘thump’ of the drum as a whole. I find that applying a straightforward low-mid EQ boost to a drum mic tends to bring with it undesirable side-effects; better, rather, to apply a momentary boost each time the drum is hit, to add weight to the attack without over-emphasising muddy frequencies in the decay or, worse still, in any spill that’s been captured. Again, this is easily done using a plug-in like Pro-MB, in this case with a single band centred at 160Hz and configured as an expander. One of the nice things about Pro-MB is that each band’s key signal can be filtered independently of that band’s own centre frequency and bandwidth. In this case, that allowed me to have my low-mid boost triggered by energy in the 1-12 kHz region, thus ensuring that expansion was activated only by snare hits and not by kick-drum spill.
It’s amazing how stubborn sounds can be when you try to alter their character through processing, though. Even with my expander adding almost 8dB of low-mid-range boost to every snare hit, the snare sound in the mix was dominated by the crisp upper mids. I could have layered the snare with a sample, but I never like doing that at the best of times, and it seemed doubly inappropriate in an organic, mostly live acoustic recording. Instead I tried opening up a couple of new processing fronts. The drums in Mike’s original mix were very dry, and rightly so; but I discovered that mixing in a dark, honky patch from HOFA’s IQ-Reverb at quite a low level made the snare seem thicker and more solid, yet not obviously wet. I also sent from the snare track to a parallel bus where I used Saturn and the SansAmp PSA-1 plug-in in series to generate a boxy, saturated ‘thud’. This sounded terrible in isolation, but in context, added the missing impact in the mid range.
The Dark Side
Once I had the drums thumping to my satisfaction, I turned to the bass guitar. My aim was to achieve a thick, warm sound that would underpin everything else without poking through too much in the upper mid range or disappearing on small speakers. Both the miked track and the DI sounded a little too clean to achieve this on their own, so I began by running the DI track through Scuffham Amps’ S-Gear amp simulator. This is really intended for guitars rather than basses, and wouldn’t be ideal if I’d been after a very subby sound, but I was getting plenty of bottom end from the miked track, and the point of the amp sim was to add some ‘meat’ in the mids. Further thickness was achieved by bussing both tracks to an aux and applying compression, saturation and additional dynamic control from Sound Radix’s Drum Leveler. (It’s a funny thing with basses: when they’re played badly, they usually need lots of compression to iron them out, but when they’re played well, as in this case, compression still seems to do something nice to the sound.)
The bass had been DI’d and miked. On the left you can see the DI processing — an instance of Scuffham’s S-Gear guitar-amp emulation and a time delay to bring it in line with the miked signal. Both tracks were sent to a bass bus, on which Sound Radix’s Drum Leveler brought the dynamics under control unobtrusively.
It’s often the case that once you have a solid foundation for a mix, other things start to fall into place quite quickly, and so it proved here. The process of reshaping the drums, as described above, took quite a while, with plenty of false starts made and blind alleys explored. Once that was done, though, many elements of the multitrack just worked, almost without any processing.
Unusually, Rachel’s vocals had been recorded through a Sennheiser MD441 — a high-quality, hypercardioid moving-coil dynamic mic — instead of one of the more usual capacitor mic suspects. The raw vocal track sounded great, but erred on the side of being too dark and rich. That’s fine with me: somehow it’s always easier to brighten up a slightly dull-sounding vocal track than to tame an over-bright, sibilant recording. Mike and the band had clearly done the same in their mix, and perhaps in order to keep up with the bright-sounding drums, had pushed it quite a long way in this respect.
So, apart from a small notch at 4kHz to take the edge off some consonants, I decided to avoid using EQ in favour of more multiband compression. When the vocal sound needs to remain natural, I find this is often a better option, because you can set it to act only when needed, rather than applying a blanket tonal change. In this case, I used two bands. One was set to tame everything below 650Hz, and the other was centred at 2.2kHz. By default, this second band applied a broad boost of 3dB or so, but it was set to compress by the same amount on the louder notes, where the upper mids were naturally more strident and didn’t need accentuating.
A Mute Point
There are times when you want to create the impression of a powerful band by mixing the vocal almost as quietly as it’ll go while still remaining intelligible. At other times, it feels right to use the vocal as a playground for creative effects such as filtering, distortion, tempo-sync’ed delays, backwards reverbs and so on. Both from the original mix and from Mike’s comments about the band’s preferences (see box) it was clear that this wasn’t either of those times! In this case, the vocal needed to be the biggest thing in the mix; yet it also needed to remain largely natural-sounding. One way to create this impression is, of course, to make the vocal really loud. However, there’s only so far you can go before the vocal stops feeling as though it belongs to the mix.
So, if filling up the entire stereo panorama with vocal delays, reverbs and whatnot is also not an option, what can you do? Well, if you’ve already made the vocal track as big as it can be, one option is to make everything else smaller!
So, having already decided that the drums worked best being almost in mono, I set about applying similar principles to the rest of the sources. A lot of engineers whose work I admire advocate so-called ‘LCR panning’. The idea of this is that you avoid using ‘in-between’ pan positions, the result being that every mono track in your mix is panned either hard left, hard right, or straight down the middle. The obvious concern with this technique is that you risk ending up with a lopsided mix, because quite often, a source you pan one way won’t have an obvious counterpart that you can put on the other side. In practice, though, I find this worry usually turns out to be largely theoretical. If you can get the balance right, the occasional bit of space on one side or the other is often a good thing; and sticking to LCR principles, combined with frequent auditioning in mono, makes it easier to achieve that correct balance in the first place.
Double Or Quits
In the ‘Breaking Branches’ multitrack, all of the instruments were represented by at least two tracks. For instance, two acoustic guitar parts had been recorded. Both were played through an amplifier, and separate mics had been placed in front of the guitar and amp. This sort of arrangement lets you really fill out the soundfield if you wish; by panning each part’s amp and mic tracks to opposite sides, you can create a huge sound. However, in this case, a huge sound was exactly what I wasn’t after, so I muted the amp tracks altogether, and cut some low-mids from the mic tracks to make them light and airy rather than dense and thick.
The 12-string electric guitar, meanwhile, had been both DI’d and recorded through a Vox AC15. Again, as if to try to compete with the drums, the amped sound was a bit shrill at 2kHz and above. I toyed with the idea of combining it with an amp simulator on the DI’d signal, but in the end, I found a combination of EQ and multiband compression that enabled me to dip the offending upper mids enough to make the amped track sit peacefully on its own. And when (with the help of some entirely correct feedback from Mike) I’d found the right level to place the 12-string in the mix, having a mono electric guitar panned hard left didn’t sound at all strange or unbalanced.
I pushed the LCR aesthetic even further in the case of the vibraphone. Layered ‘high’ and ‘low’ vibes parts had been recorded very nicely in stereo; but I didn’t want the vibes to have stereo width, so in each case I simply muted one half of the track. This had the added advantage that by muting the high part’s low mic and vice versa, I could create more of a tonal distinction between the two parts. As Mike noted, the vibes stand out more in my remix than in the original, perhaps in part because there’s more space for them to be heard.
The general idea — which is, perhaps, the point of LCR mixing — was to treat every instrument as a ‘point source’ within the mix, with a well-defined position either in the middle of the panorama or at its edges, and no stereo width at all. What’s more, although I applied equalisation, some compression and various ‘warming’ treatments to all of the above-mentioned elements within the mix, none of them got any reverb or stereo effects.
The LCR approach to panning in this mix, which allowed Sam to drop some of the instrument doubles, or second mics, left him plenty of space in which to make the vocal seem ‘big’ by comparison, an effect that was enhanced with an Eventide H3000 Factory widening patch.
When this approach works, it gives you a mix that is naturally very wide and full, but where every instrument is clearly defined. (This sort of clarity can be as much a curse as a blessing where the performance or recording are not up to the mark, though!) And, to return to the point I abandoned a few paragraphs ago, in this mix, it meant that a modest amount of reverb and doubling from an Eventide Harmonizer plug-in were enough to make the lead vocal stand out as bigger, wider and fuller than everything else, without coming across as obviously effected or unnatural.
On The Buses
The purist in me rebels against the idea of mix-bus processing, but usually loses out to the pragmatist who points out that something needs to be done about the 200Hz region. In this case, I discovered that I could make the bass and low mids more consistent by feeding the Bass Boost on SoundToys’ Radiator plug-in into a multi-band compressor, then using an EQ to apply a gentle shelving cut. Finally, I often find that I like the effect of mix-bus compression more on sparse, downtempo acoustic tracks like ‘Breaking Branches’ than on faster or busier material, and a couple of dB of very gentle gain reduction from IK Multimedia’s LA-2A emulation worked nicely (all of these plug-ins were running in series as inserts on the master bus).
After an initial round of mix feedback, where Mike quite rightly pointed out some balance issues, the rest was detail. Mike had created a very nice swirly delay effect for parts of the electric guitar track; I hadn’t tried to recreate it in my mix, partly because it violated my self-imposed ‘no stereo’ rule and partly because it sounded quite complicated. Having heard my initial mix, Mike bounced this effect out as a stereo audio file and suggested I try reinstating it, and it did indeed prove to be the missing piece of the puzzle.
A Pleasant Change
As I mentioned at the start of this article, the usual focus in Mix Rescue is in making the best of a bad job. By contrast, this project was more about continuing someone else’s good work, which is much more representative of the normal role of the mix engineer. The recording was good, the performance excellent, and most importantly, all the crucial production decisions had been taken at the right time. Nothing needed tuning, timing correction or re-recording, and I wasn’t left in the awkward position of needing to change the arrangement after the fact, or impose a ‘sound’ on a band who lacked their own. Simply putting up all the faders would have produced a more than listenable track; and inasumuch as I was able to build on what Mike had captured, I was really just supplying the icing on a very tasty cake. And that’s exactly the way it should be!
Share & Share Alike?
Mike Stoodley’s relationship with the Broken Heartbreakers is a much closer one than most recording engineers enjoy with their clients. He used to play bass in the band, and is clearly a trusted collaborator whose creative input is valued. This, it seems, suited the musicians, who were keen to remain closely involved with all aspects of the production process, including the mixing. Geographical differences made it impossible for everyone to mix together, but they achieved the next best thing by exactly duplicating the mix setup in different cities. This meant that Mike and the band could not only share draft mixes, but Reaper project files. There are obvious advantages to this sort of arrangement, but one or two limitations and pitfalls, too. For example, they were restricted to working with free or very affordable plug-ins, as everything had to be installed on two different machines. That’s perhaps not such a big problem, as there is plenty of quality freeware and shareware out there.
What perhaps was a more subtle challenge was the inevitable tendency for a band in that position to judge with their eyes as well as their ears. The Broken Heartbreakers, for instance, are very cautious about compressors affecting the integrity of their recordings and prefer not to use them at all. As we all know, though, there’s compression and compression. It’s one thing to hear a compressor pumping or distorting and dislike the audible result; quite another to open up a mix window in your DAW, notice that someone has inserted a compressor and convince yourself that it must be doing something evil. Compression was invented to spare human beings from having to constantly move faders: used sensitively, it simply automates level adjustments that the mix engineer would otherwise make by hand, and need not sound any less transparent or natural.
Unlike Mike, I was free to serve up a complete mix without having to make the ingredients available for inspection. By the time I’d finished, I was using six instances of IK Multimedia’s White 2A; like the classic LA-2A that it emulates, this is a supremely smooth compressor, and up to a point, it almost certainly controls signal level more invisibly than I could achieve just by moving faders. I also leaned fairly heavily on Sound Radix’s clever Drum Leveler plug-in, which helped me to flatten out the kick, snare and bass tracks without imposing a ‘compressed’ sound.
One other quality I like about the White 2A plug-in — and, indeed, some other vintage compressor emulations — is that although you can’t hear the gain reduction at work, it does have a clear effect on the tonality of the sound passing through it, somehow cleaning up the low mids and adding density further up the mid range. To reinforce this shift, I was also liberal with my use of plug-ins like Slate’s Virtual Tape Machines and SoundToys’ Radiator. The effect of any one of these plug-ins was marginal, but taken together, they made a big difference to the feel and presence of the mix as a whole. The Broken Heartbreakers liked the results, and didn’t complain about the added hiss, but I do wonder whether these plug-ins would have survived if I’d been sharing my mix files with the band as I went along!
What Gives?
The psychology of mixing can be perverse. There are times when you labour for hours over something, only to find that no-one notices; but there are also times when everyone hears an improvement even though you feel like you’ve done nothing to bring it about. Some of the feedback I got from Mike and the band definitely fell into the latter category! For instance, Mike himself had found it hard to get the acoustic guitars to sound as he wanted them, and had employed quite a lot of processing; whereas all I did was dial in a fairly unremarkable EQ curve and add a little compression and saturation. Likewise, he asked what I’d done to increase the excitement in the song’s instrumental section when, apart from a few small fader moves, I’d done nothing. Mike was curious, too, to know how I’d introduced a sense of smoothness into the mix as a whole, but I had to admit I’d never even thought about this as an issue.
I think the most likely explanation is that all these improvements are probably real, but they came about as side-effects of more general work on the nuts and bolts of the mix. Once I’d reworked the drum sound and put into practice the LCR panning described in the main text, the other benefits came for free: acoustic guitars and vibes had room to breathe, the band’s natural dynamics in the instrumental part came across more directly, and the vocal felt big and up-front yet still integrated into the mix. Mike was quick to appreciate the benefits of the LCR approach once he’d heard the results, saying, “My immediate reaction to this was that you had worked some dark magic because it didn’t seem unstable or lurch around, despite being obviously ‘unbalanced’.” And, on the smoothness front, the darker drum sound in the remix perhaps meant other things didn’t have to push so hard in the upper mids to compete.
Remix Reaction
Mike Stoodley: “I think [guitarist] John summed up the difference rather nicely: he said ‘I can imagine Sam’s mix on the radio in way that I can’t imagine ours,’ and ‘Sam’s mix sounds “professional” whereas ours sounds like a really, really good demo.’ Now whether you prefer a “professional” sound to a slightly rougher one is a highly individual call. I’m certainly not thinking any less of our mix — we are all still stoked with it. I think Sam’s mix is better because it has everything that ours has plus some fairy dust, and most importantly, it passed the goosebump test.
“He made some key decisions that opened up the sound a lot more than what we were able to do. With a bit more skill and experience, we could have made those calls too, but didn’t. Mixing is an accumulation of small decisions and the technical execution of them; he made better decisions, and executed them better, than we were able to. It’s kinda like cricket... they batted better, bowled better, and fielded better... Another way of looking at it was that we had a 90 percent completed jigsaw, he came along and went ‘Oh that bit goes there, this is upside down, these blocks are OK but they join together with these pieces, and that edge actually goes on the other side.’ And bingo, our nearly done jigsaw is done.”
Audio Examples
This month’s Mix Rescue was unusual in that we began with a good mix of a good recording! I’ve placed a number of audio examples on the SOS web site, in which you can hear how the raw sound of the various tracks was enhanced or developed at the mix.