Mix Rescue: Ollie Wright

The main quadruple-tracked rhythm guitar part was another sticking point, because Ollie had used almost identical sounds for each layer. While layered guitars are certainly par for the course with this kind of production, if you use the same sound for each layer you’ll find that the sound quickly becomes stodgy and bland, with very little character. He’d also made the all-too-common mistake of overdriving his guitars too far, so the power and harmony of the parts was getting lost in a hash of high-mid frequencies.

A further problem with the guitar parts (and, indeed, the bass) was that their timing wasn’t tight enough against the drums. There are some styles of music where small timing variations between performers are a vital part of the appeal, but this isn’t really one of them! If the bass doesn’t agree with the drums, the apparent power and momentum of the track as a whole suffers, and the same goes for the rhythm guitars. You’ll also find, with guitar parts, that the attack of each chord will never sound punchy unless the different layers agree on when it occurs.

Timing was a concern with the lead vocals too, especially in the choruses, which were again quadruple-tracked, but tuning was also an issue. No matter how good your mixing chops, you won’t be able to get the vocal to blend with the backing track in the way it does on this kind of commercial record if consonants are flamming against each other everywhere. Fairly tight tuning is also characteristic of this mix style, and is partly what enables the smooth, slightly chorused sound that many listeners now expect.

The vocals had been recorded using an SE Electronics SE2200 microphone, which seems to be a very common budget choice these days. What I’ve noticed over time, though, is that the mic’s hard, bright tone tends to suit softer and breathier singing much better than full-on tonsil-battering, and this song was a case in point: while the verse and middle-section vocals were pretty respectable, the mic was unpleasantly emphasising a hard edge that (quite naturally) appeared in Ollie’s voice when he shifted his larynx into ‘arena’ mode!

Overall, then, what was primarily holding back this song wasn’t Ollie’s mixing as much as his programming, initial sound choices, editing, and vocal-mic choice — in other words, the production work prior to the mixing stage. Indeed, Andy Wallace’s mixing approach is, by all accounts, pretty simple (as was mine for this track once I’d spent some time tackling the underlying production issues). Given that many mixes sent in to SOS also suffer from lack of attention to pre-mixdown details like this, I’m going to go into a bit more depth about how you might approach preparing a project such as this for mixdown.

The way I usually start is by locking the bass pretty tightly to the drums, using the song’s metric grid as a visual guide. I say ‘guide’ here advisedly, as even with completely quantised drums you won’t get the best subjective timing match by just using your eyes. For a start, it’s not uncommon for a specific bass note to look fine even though you’re actually hearing it as out of time, and you need to trust your ears over your eyes. However, by the same token, a note that actually looks and sounds in time in isolation can feel like it’s early if the preceding notes have been progressively dragging, and vice versa. I also often find that the bass/kick combination drives along better for me if the kick leads just fractionally — although this kind of decision comes down more to personal taste than anything.

It’s worth pointing out that after you’ve done a section of detailed timing edits (for instance, in this case there were almost 50 edits in the bass part for the main riff section), you’ll typically have lost a certain amount of perspective on how the timing of the whole section feels, so it’s important to listen to the section from the top again and give yourself the opportunity to refine more subtle medium-term timing drifts that aren’t as apparent when you’ve got individual notes under the microscope.

Guitars were next on the surgeon’s table, and while the editing process here was fundamentally very similar to that for the bass, there are some additional considerations when dealing with panned double-tracked parts like these. The first thing is that the timing between the different opposition-panned guitar layers may need to be even tighter than that between the guitars and the bass/drums, because the brain is extremely sensitive to timing disparities between similar sounds arriving from different directions — even slight flamming between notes/strums can get distracting. The second thing to say is that you’ll probably get good results quicker if you deal with micro-level timing corrections for the guitar layers one at a time. This will be most time-consuming for the first part, but you can then use the first edited part as a visual reference to speed up editing of the remaining layers. That said, with layered parts, I find it easier to edit each one in turn without hearing the others — I can hear what I’m doing better — and only check their combined effect once they’re all edited. At that point I may find the odd strum here and there that needs shifting a little across all parts, but there’s rarely very much left to do otherwise.

You’d be well advised to get to know your sequencer’s keyboard shortcuts if you’re embarking on this kind of work, because it can make an enormous difference to your work rate. On a commercial level, one of the reasons for the rise in importance of professional Pro Tools operators is that this kind of editing is so widespread and these guys have such an instinctive grasp of all the system shortcuts that they can deal with repetitive editing tasks in a tenth of the time it would take mere mortals!

Once the MIDI files were imported into the project, I opened up Sonic Reality’s Ocean Way Drums (running as a VST instrument in Reaper, using Native Instruments’ Kontakt Player) and surfed through the sounds until I latched onto a snare and a couple of crash cymbals that sounded like they might work well with the rest of the arrangement. Triggering these sounds from Ollie’s MIDI files, I tuned them into line with the song’s key and then balanced the different virtual mic signals to add in the right amount of room sound. A lot of people ignore tuning their drums to match the song key, but it can have serious practical advantages. The lower pitched frequencies of a drum are often what give it its sense of power, but they can also be difficult to blend into the mix. If you tune these elements to the track, they’ll blend much better, so you can have a more powerful snare with more low-end.

The remainder of the drum work was to go through the MIDI data and adjust Ollie’s velocity programming. For the snare backbeats in the chorus, I left the backbeats fairly consistently at the maximum 127 velocity level (in line with the all-out rock explosion going on at that point), but made sure that the other hits were scaled down appropriately, to give them a little less prominence and help the groove to work more smoothly. A good example of the effects of these changes can be heard during the snare fills in the loud instrumental section following the chorus vocals. Originally, all those hits had a similar dynamic level and therefore a rigid ‘machine gun’ feel, whereas a bit of a velocity ramp through each fill immediately made things much more musical. Another trick you can use with a lot of the big multisampled drum sample libraries is to alternate between the supplied left- and right-hand stick hits during fast snare fills, and that helped in this example.

For the cymbals, I noticed that although there was a certain amount of velocity variation in Ollie’s original programming, this variation wasn’t necessarily supporting the music. While random velocity variations are one characteristic aspect of human performance, much more important are the instinctive and intentional musical stresses. For instance, it’s usually fairly natural to put most stress on the main downbeat of a 4/4 bar, a little less on beat three, a little less on beats two and four, and even less on successively smaller subdivisions of the main beats. However, while this can be a useful rule of thumb, you do need to listen to the track and think where the musical stresses are in context. In this case, each of the main guitar onslaught’s three eight-bar subsections had a slightly different stress pattern: more stress on beat four in the first subsection; a fairly regular stress pattern underneath the vocals; and increased weight on beats two and four during the third subsection.

I’m usually a bit hesitant to completely replace someone’s drum parts, because there’s usually some aspect of the original sound that appealed in the first place, so I also asked Ollie to re-bounce stereo versions of his own snare and cymbal parts for layering behind Ocean Way Drums. With Ollie’s parts mixed in at a lower level, I was able to retain more of a connection to his original sound, although I did find that I much preferred the attack of the Ocean Way Drums snare, so I heavily limited Ollie’s snare to extract just its sustain.

In the first instance, I wanted to increase the sense of pitch, as well as giving the sound some edge through the hash. For this I turned to a couple of the Fender amp models (Tweed and Tweedman) in Native Instruments’ Guitar Rig 3, being careful to keep the gain fairly restrained, to avoid anything too ragged. I’m no guitarist myself, so I can’t say I had any real plot while fiddling with the amp controls, but the main thing is just to concentrate on how the resultant sound complements what’s already there.

While the additional Guitar Rig 3 parts helped the overall tone, I still felt a need for some extra fullness and width, so decided to add in another double-tracked re-amp layer for that purpose. I’ve found that different layers of amp-modelled guitar seem to work better together when you don’t use the same modelling for them all, so rather than firing up Guitar Rig 3 again, I pulled out some handy freeware plug-ins from Simul Analog instead: the Oberheim PS1 phaser, the Boss DS1 distortion, and the amp-modelling Rednef Twin. (Do you think that last one might have been reverse engineered?) A bit more blind tweaking of controls and I had something that fitted the bill.

For the bass, I was happy to work with the amp-modelled track that Ollie had supplied, although I did compress it heavily, to keep it really solid in the track against the wall of guitars, using a combination of 3-6dB slower 2:1 compression from Bootsy’s Density and faster 8:1 compression from Stillwell Audio’s The Rocket. The latter plug-in’s fast time-constants can really distort bass instruments if you’re not careful, and while this isn’t necessary a bad thing in some cases, you should probably still tread carefully. As I frequently do, I also layered in a sub-bass tone using Reaper’s simple built-in Reasynth instrument to gently fill out the lower octaves of the mix and make it feel a little more ‘hi-fi’. What was tricky, though, was deciding how much of the mixdown’s low-mid range was going to come from the bass part, and how much from the guitars.

Although I tend instinctively to rely more on the bass for low end in my mixes in general, high-pass filtering any conflicting guitar low frequencies out of the way, some careful comparison of my mix in progress with the Biffy Clyro album indicated that a fuller low end to the guitar sound was going to be more appropriate, spreading the low-mid range across the stereo image and increasing the sense of size during the chorus sections. However, there simply wasn’t enough room in the mix for the guitars and bass to let it all hang out in this frequency region: it just made the overall mix tone muddy. So what I ended up doing was filtering out less LF from the guitar parts and then cutting the bass by a couple of decibels at both 220Hz and 360Hz, using Universal Audio’s Neve 1073 EQ plug-in for their UAD1 and UAD2 DSP cards.

The chorus vocals could afford to be tightly pitched most of the time, because I wanted them to have that characteristic close-double-tracked sound which doesn’t take up much space in the mix, so most notes found themselves adjusted to some extent. However, for the verses, and especially the middle section, I reverted to my more usual ‘hunter killer’ method, correcting only the bare minimum of notes required to keep the tuning secure. In any given phrase, you’ll usually find that some notes are more important to the sense of tuning than others, so if you focus on those you’ll find that the less important pitch vagaries on other notes keep a whiff of realism about the performance as a whole. The more uncorrected notes I feel I can get away with in an exposed lead vocal, the happier I am, and although there are parts of the lead vocal in my remix that are clearly still technically out of tune, I’d not have left them that way if I didn’t feel that they improved the emotional impact.

Timing correction also played an important role for the vocals, especially during the choruses, but there wasn’t much difference in my approach compared with when I was editing the guitars: I worked through one vocal layer at a time, before combining them and making final adjustments. The moment the vocals were in tune and in time, they blended better with each other and with the mix. Also, although I’d initially suspected that sibilance might be a problem, this sorted itself out once it was tidied across all the parts.

The verse and chorus vocals were processed separately. For the former, I turned to the UAD emulation of the LA3A, which provides lots of smooth dynamics control, but with a clearer tone than that of its more famous valve sibling, the LA2A. For the choruses, I used The Rocket again, for its general attitude and fast time-constants, although with hindsight something less edgy might have worked better, because I later found myself fighting with the harshness of the SE2200’s sound, and I reckon that the choice of compressor probably exacerbated the situation a bit. The difficulty I had was that although I could fairly easily find the problem frequencies in the vocal recording (3.8kHz and 5.4kHz), the singing only became unduly harsh on certain notes, so simple EQ dips were not the answer. My solution was to dial up the Platinumears IQ4 dynamic EQ plug-in to target these frequency areas only when they got beyond a certain level. Each EQ band in IQ4 can act a bit like a compressor, so you can set a threshold beyond which the EQ begins to reduce its gain. The trick in my case was to set the thresholds of my two EQ bands so that gain reduction was triggered mostly during harsh-sounding notes. I don’t often resort to this kind of processor, partly because it can be fiddly to set up, but on occasion the precision effects of dynamic EQ can be just what the doctor ordered.