Sometimes, simply taking the time to identify and address a couple of main obstacles before you get busy with the effects can be enough to unleash a mix's potential...
The regular Playback column in SOS is (next to Inside Track) always my first stop the moment my copy of the magazine hits the doormat, and following reviewer Nell McLeod's favourable write‑up of an EP from a band called Phre The Eon, I thought I'd check out their MySpace page. However, much as I enthusiastically agreed with Nell's opinions about what they were doing musically, I couldn't help feeling that the band were being let down by their mix values, especially on a great indie/funk hybrid number of theirs called 'Everybody's Falling Apart'. So I decided to get in touch with the band and find out if they'd be up for the Mix Rescue treatment. Fortunately, they were, and their engineer Warren Hilton (another SOS reader) promptly delivered all the multitrack files I needed to get to work.
The rhythmic groove of the chorus seemed to me to be crucial to the success of this track, so I began work with the drums and bass foundations. The drum parts had been created using the DW kit in FXPansion's BFD, and I imagine that something like a Roland V‑Drums system had been used to create the part, because the performance was excellent. However, while the hi‑hat, snare, tom and cymbal close‑mic multitrack output files were completely without any spill, the kick, under‑snare and overhead tracks contained a level of simulated hi‑hat spill that stymied my attempts to create a decent kit balance — as soon as I faded up any of those tracks to where I wanted them, they made the hi‑hat sound splashy and horrible. The situation was, in fact, worse than I'd expect with a hat‑heavy live kit recording, because the different spill signals in BFD seemed to be phasing in a nasty way that I've never encountered with any real drum session.
One of my first jobs, then, was to try to rebalance this aspect of the sound. The kick sometimes played at the same time as the hat, so EQ was a better solution than dynamics there, and a 4dB shelving cut above 1kHz made short work of the problem, without compromising the kick's own nice, sharp attack. However, for both the other leakage situations, side‑chain‑triggered dynamics proved to be the order of the day. For the under‑snare track, I loaded in a gate plug‑in, feeding its side‑chain from the main over‑snare signal. Because the latter had no spill on it, getting the gate to trigger correctly was a doddle, and all that I really needed to do was adjust the gating envelope to taste with the plug‑in's Attack, Hold and Release controls. For the overhead processing, on the other hand, I ended up using a compressor controlled via its side‑chain from the virtual hi‑hat mic, a setup that ducked the overheads level every time the hi‑hat played. Although setting the threshold was again easy in this case, the ratio and time‑constant controls took a little more tweaking to achieve the best hi‑hat spill reduction without introducing distractingly overt gain‑pumping side‑effects. In the event, I was able to pull back the spill by a useful 3dB before things became too unnatural, which was enough to bring about a more suitable balance with the cymbals.
After that bit of troubleshooting, I was able to concentrate more on sculpting the kit tone as a whole. The two virtual kick mics covered most of the bases already, so there was little to do there: for the 'inside' mic, I just used DDMF's LP10 EQ to pull down one trailing resonance with a narrow cut at 97Hz; and for the 'outside' mic, I adjusted its phase‑match using Betabug's Phasebug and gave it a bit more punch with Voxengo's Transmodder. This last plug‑in is fast becoming one of my first‑call choices for drum processing, because of the way it enables you to focus transient enhancement effects into specific frequency regions. It takes a bit of setting up, but it usually rewards the effort amply!
In this case, I first set up one of Transmodder's internal trigger channels in low‑pass mode with a cutoff frequency of 100Hz, in order to get it to respond only to kick‑drum hits, rather than to any of the track's simulated spill. Then I set two of the plug‑in's transient‑driven EQ peaks to enhance the impact in one‑octave regions centred at 55Hz and 390Hz. Although each of these bands was ostensibly set to boost by 10dB, the actual gain change shown in the EQ curve display was more like 5‑6dB and the effect in the mix was only moderate, as you can hear from the audio files accompanying this article.
The snare tracks didn't need much processing either. There were three mid‑range resonances that I wasn't tremendously keen on, but narrow 4‑6dB cuts at 342Hz, 497Hz, and 645Hz from another instance of LP10 swiftly dispatched those. Then I evened out the performance slightly, using the Tin Brooke Tales freeware TLS Mastering Limiter, before adding a gentle 2dB peaking boost at 190Hz from the Cockos ReaEQ, to warm things up a touch.
The only real problem remaining after those adjustments was that the level of the snare's room sound in the overheads was quite low, so the snare felt dislocated from the kit until I added some artificial reverb from SSL's new X‑Verb (running on their Duende processing platform). Finding a suitable X‑Verb patch involved a fair bit of surfing through the presets in the first instance, and then some extensive tweaking by ear — turning off the early reflections, reducing the reverb time to around half a second, and then just trawling through the remaining parameters to see whether adjusting them either way helped improve the snare's tone or blend. While some engineers get a bit sniffy about using presets, reverb algorithms are often so complicated these days that I find it much more efficient spending time locating a good preset to tweak than burning the midnight oil setting up dozens of inscrutible and interacting algorithm parameters from scratch.
I treated the overheads to some fairly fast 2:1 compression from DDMF's new NY Compressor to improve the cymbal and snare sustain and add a bit of the kind of gain‑pumping action that most bands seem to expect nowadays. Gain‑reduction readings of around 5‑7dB seemed to work well, in combination with an 85ms release time, so I turned my attention to the overall tone. Although the triggered ducking I described above had sorted out the hi‑hat's overall balance, the fizziness of its room sound still seemed overbearing, so I tackled this with a 5dB shelving cut at 13.5kHz from Sonnox Oxford EQ. A couple of 3dB mid-range peaking boosts (a broad one at 250Hz and a narrower one at 1.7kHz) also helped bring the body of the snare and cymbal sounds more to the front of the mix.
Where I'm after a punchy kick‑drum sound, I often filter out the low end of the overheads so that the kick's close mic dominates — but at the outset of this particular mix, the low end of the room sound seemed to have something to offer, so I left it unfiltered, albeit tightened up slightly by an instance of Flux Alchemist expanding the dynamic range below 150Hz. By the time the whole arrangement was up and running, though, a drier low end seemed to work better, so I returned to the Oxford EQ and high‑pass filtered the overheads at 106Hz instead.
The hi‑hat close mic had a funny low‑frequency resonance at 73Hz that called for notching with another LP10, but otherwise my processing on this track was mostly concerned with trying to make it tighter and clearer: the raw audio was, again, hissy and somehow distant, which meant that the close-mic track failed to increase the instrument's definition when mixed in with the overheads. A 6dB peaking boost from LP10 at 3.9kHz immediately brought the sound forward (as any emphasis of the upper mid-range 'presence' region usually will), while SPL's Transient Designer sharpened up the stick attack to help drive the overall rhythm along. An instance of Voxengo's PHA979 phase adjuster also upped the instrument's mix solidity one more notch, by improving its combination with the remaining overhead hi‑hat spill.
It may be a weakness of mine, but I can never get very worked up about tom close‑mic tracks! This mix was a case in point, since I did very little processing to them at all. Some limited‑range gating from SSL's Drumstrip plug‑in was handy to stop them ringing on too long, and a little of the snare's reverb improved the blend, but that was it really. That said, I did end up riding the levels of these tracks at a later stage, to get the balance right for each fill. As for the cymbals, these actually seemed to be coming through pretty nicely from the overheads, so I didn't use a tremendous amount of them, low‑pass filtering at 5.8kHz with Oxford EQ and shaving off stick attack with Oxford Transmod, just so that I could add a fraction more body to the cymbal tone.
The bass sound from the band's original mix had been constructed from a combination of DI and miked‑amp tracks. The DI was pretty usable once I'd opened up Flux Epure II to filter out some harshness above 3.5kHz and tame a strong resonance at 166Hz, and I only felt the need to warm it up a bit with a further broad 2dB peaking boost at 350Hz, and a gentle 2‑3dB of levelling from Universal Audio's LA2A plug‑in, running on a UAD2 card.
The miked amp track, however, didn't really complement the DI in the way I'd have hoped, as there wasn't really much character or fullness to it. I tried time‑aligning the waveforms on the two tracks (which often helps) and messed around with EQ for a little while, in an effort to draw out something attractive, but eventually just canned the track and ran the DI signal through IK Multimedia's Ampeg SVX instead. This software's '60s Portaflex emulation turned out to be my favoured option in this instance, although I did move the virtual mic back to its 'Far' position, to make it a bit rounder sounding.
The amp modelling helped a lot, and the track didn't even seem to need any phase matching to get a good result (which is unusual, in my experience), but I still hankered after more low‑end power for this particular song. I wanted the bass to make a real statement on this track, but it felt a bit 'slimline' as it was, so I decided to add some artificial low‑end definition by programming in an additional surreptitious sub‑synth part. Andreas Ersson's little freeware Iblit instrument stepped into the breach on this occasion, providing a single‑oscillator, square‑wave patch, which I low‑pass filtered fairly severely (but with low resonance) to remove unnecessary upper harmonics in a smooth and progressive way. That worked a treat, and gave me the basis upon which I could begin building up the remainder of the arrangement.
Before I started adding in the guitars, I figured it was time to get the lead vocals into the balance, so that I wouldn't forget to leave them space in the final texture. This was one area of the track that was comparatively plain sailing, requiring only straightforward processing with SSL's Vocalstrip and Cockos ReaEQ. In Vocalstrip, a 95Hz high‑pass filter kept the low end clear; a couple of decibels of peaking cut at 7.2kHz scotched an element of harshness that, I expect, was derived from the inexpensive LD Systems D1122 condenser used for recording; a moderate 3:1‑ratio soft‑knee compression setting (attack 10ms, release 40ms) reduced the dynamic range; and a de‑esser compensated for the compressor's inevitable emphasis of sibilants. ReaEQ refined the low‑end tone a little further, raising the 800Hz and sub‑250Hz regions to give the voice more size and authority.
The acoustic guitar was next, this instrument having been recorded using a matched pair of AKG C451s, although a DI signal had also been taken, to give three tracks in total. All of these tracks had rampant sub‑bass on them, so I had to high‑pass filter them all before I could audition them at any volume without bidding a tearful farewell to my subwoofer!
Once I was able to concentrate closely on the tracks, I felt that they'd all captured a sound that was too abrasive and percussive, so I loaded in another Oxford EQ and lifted a broad region centred on 1.1kHz by a couple of decibels, to bring out more of the note pitches. The low‑end thump was also reduced with a further 60Hz high‑pass filter and a small (1dB) peaking dip at around 315Hz.
A track of fairly clean electric guitar, comprising quarter‑note rhythmic chords, was rather veiled — so I used Flux's Epure II to high‑pass filter a small amount of low‑end mush and to bring out some body at 540Hz and detail above 4kHz, although the boosts in each case were only of the order of 3dB. This helped a certain amount, but some recorded reverb (presumably from the amp emulation used for recording) still left the track feeling bogged down, so I decided to gate out the reverb tails between the chords. In the first instance, I did this by triggering the gate from the chords themselves — but I found it difficult getting the gate to trigger reliably enough on all the chords, so I switched to triggering it from a specially created quarter‑note trigger track. (Naturally, I didn't feed this track to the mix buss itself: otherwise I'd have ended up with a house remix!)
The two lead‑guitar lines both felt rather limp and pudgy, but some experiments with EQ didn't yield the kind of extra hardness and clarity I felt would help the parts make their presence felt in the arrangement. Distortion is often a good medicine for this kind of ailment, and Mokafix Noamp was soon drafted in to provide one of my favourite flavours in this department (a simulation of Tech 21's Sansamp), which meant that I only then needed some high‑pass filtering and a few decibels of fairly narrow-band cut at 280Hz to arrive at the final mix tone.
Although I was still just working on the chorus section, all its tracks were now in place, so I could strap on a fairly run‑of‑the‑mill 3‑4dB of fast 2:1 master‑buss compression from SSL's Stereo Buss Compressor and then step back to take stock of how the overall production was stacking up. While the mix processing so far had clearly helped remove some clutter, it was clear that there was quite a way further to go in terms of achieving a big, satisfying sound, even taking into account the kind of contribution I've learned to expect from normal mix‑send effects, which I'd yet to involve.
The first thing that hit me was that the timing and tuning of many of the parts was compromising both the punchiness and the blend overall, so I decided to test this hypothesis with a bout of audio editing. As the drums felt good to me, I decided to use their performance as a reference point for the timing, and took the scissors to all the other parts to lock them in more closely. As I'd hoped, this improved the sense of punch and momentum significantly.
The band had delivered a pitch‑corrected vocal part with their files, and while the processing had been done pretty sensitively, I still felt that I might get a better result from re-doing it in Melodyne Editor. That software's polyphonic tuning‑correction also came into its own for the acoustic guitar part, which seemed to be undermining the song's harmony in its untreated form. However, this was one of those occasions with Melodyne where I was in two minds as to whether the chorus‑like side‑effects of the pitch processing were too high a price to pay.
The guitar definitely seemed to wash out after correction, losing a lot of transient definition — so much so that I felt moved to add an SPL Transient Designer plug‑in, to try to compensate to some extent. In the end, though, while I considered the cure to be just about better than the disease during the choruses, I resolved to put up with the tuning issues during the rest of the song, wherever the arrangement exposed more processing artifacts than I cared to hear.
Even with tuning and timing tucked in more snugly, the main quarter‑note rhythm‑guitar part didn't really seem to fill out the chorus sound properly, so I took some further steps to fatten that out. Firstly, I was able to edit different sections of the part onto another track, to conjure up a double track, which could then be panned to the opposite side of the stereo image, increasing the width of the chorus without imbalancing the picture to one side. Then I scoured through some of the other guitar tracks elsewhere in the song and found a fuzz‑guitar track from the pre‑chorus that happened to fit the chorus too. This was edited into a double‑tracked part, as before, and lightly sculpted with an Oxford EQ plug‑in (+1.5dB at 1.4kHz and ‑1dB above 4kHz), but then Bob was definitely my uncle!
At last, the chorus began to feel like it was really holding its own, so I began to pursue the final touches of blend and sheen that global send‑effects can offer. It often pays to wait until late in the game before adding such effects, because it forces you not to skimp on vital processing and arrangement work. If you start on the send effects before you're excited about the dry mix sound, you're going to over-compensate for the lack of excitement by adding more effects than you need. Because I didn't rush the processing and arrangement stage, my use of send effects could actually afford to be pretty minimal, which meant that the overall sonics were able to remain clearer and more up-front than in the band's original version.
There were three main effects, all of which were very much 'standard issue': a stereo widener (my usual short, pitch‑shifted delay patch), a stereo quarter‑note delay (with a splash of feedback and some gentle high‑pass filtering), and a mid-range focused stereo ambience reverb from SSL's X‑Verb (comprising a 0.4s reverb tail with 8ms of pre‑delay). The widener was mostly for the vocals, while the delay helped to add sustain to the vocals, drum overheads and guitar solo. The ambience reverb, on the other hand, was the most general‑purpose of the treatments, applied to almost everything to some extent, in search of a more cohesive overall blend. The drums and guitars had a little more reverb level in general, to keep them behind the vocal, and I even added some reverb to the DI element of the bass-guitar part (not something I do that often) to help that gel with the other parts, given its unusual prominence in the balance.
With the chorus in the bag, much of the rest of the mix unfolded along very similar lines. There was the tuning/timing‑correction to sort out across the other sections first of all, then a few arrangement edits to enhance the long‑term mix dynamics (for example ditching the acoustic guitars for the first verse), and then some slackening off of the processing for the vocals and some of the guitar parts in response to the extra space in the arrangement during most of the other song sections. Some ride‑cymbal stick noise in the outro needed toning down with multi‑band compression from Flux Alchemist, but that was the only remotely unusual processing decision.
There were also no additional send effects, just the usual splurge of mixer automation to balance the lead vocals, as well as a handful of other little rides on other tracks — for example, increasing the reverbs by a couple of decibels during the outro section to bump up the subjective size there. However, one well‑known trick that did come in very handy when finalising this particular mix was listening to the balance without various combinations of the drums, bass and lead vocals, because the prominence of all these tracks in this song made it difficult to judge the finer points of the other mix layers and the send effects.
Looking back over the whole remixing process, I think this month's challenge was quite typical of a lot of home‑brew productions. For a start, it's almost an unwritten rule that Mix Rescue candidates under-estimate the importance of timing, tuning and arrangement edits, unwittingly setting up hurdles for themselves that mixdown processing and effects can't adequately surmount.
However, this song has also been an example of how most problematic mixes actually only present a few central issues (in this case, reducing the hi‑hat spill and filling out the bass sound), and once these are addressed, the rest of the mix can pretty well fall right into your lap. So when you're building up a mix track by track, and you get to a point where every new track you add feels like it needs masses of work, take that as a warning from your mix that you've swept something important under the carpet earlier on in the process.
Mancunian band Phre The Eon have been performing since 2007 and comprise Gary Phethean (lead vocals, electric guitar), Gary Kirk (backing vocals, acoustic guitar), Burg (bass), and Colin Ramage (drums). Judging by their MySpace presence, their emphasis on natural musicianship and simply having a good time has endeared them to a fair few punters, and they've also elicited praise from BBC6 Music's Introducing.
Gary Phethean: "On the original mix of 'Everybody's Falling Apart', our engineer Warren Hilton did a great job of capturing the mood of the track and giving it a professional feel, but we didn't realise at the time that it didn't have the extra kick and punch it needed to really make you want to get out of your chair and dance!
"When we heard Mike's remix we were immediately impressed by the real stomp that the bass and drums now have, and by the extra boost he's given the choruses. The bass definition means you can really hear and feel the heart of the track, so the chorus has a really strong uplift and the changes in the riff patterns are clearer. The extra strength and growl given to the vocals has boosted the melody throughout and given it a nice polished professional sound.
"Overall, the new mix is much more vibrant and full, and all of us in the band are still continuing to notice smaller, more subtle changes throughout — in the edited guitar parts, harmonies, drum patterns and, well... everything else. Without doubt, it's an awesome mix. To be honest, we're a bit a gutted we can only get the Mix Rescue treatment on just one track!” www.phretheeon.com