As he gets to grips with another reader's track, our mix engineer shows that distortion can often be as useful as EQ in sculpting sound sources.
Doing a decent mix at the end of two days of arduous recording is a pretty tall order, as Berkshire-based band The Wrong'uns discovered upon listening back to the studio bounce-down of their song 'Rothko' in the cold light of day. To be fair, it was certainly gutsy, but the bass warmth and drum compression had both been hyped at the expense of the guitars and vocals, creating a sound at odds with the more organic vision they'd had in mind. As an SOS reader, guitarist Rob James was least satisfied with the outcome, so had a crack at remixing the multitracks himself. However, despite the fact that he undeniably achieved a more open sound, the band nonetheless felt that something about the drive and fullness of the original version left the remix sounding rather too much like a demo. Unsure of how to square this circle, Rob dialled the Mix Rescue hotline...
The track list was nothing if not succinct: just 14 mono tracks. The drums had left and right overheads, a mono room mic, and close mics for kick, snare and one tom; the bass was DI'd; there were three rhythm-guitar parts (two acoustic, one electric), and two further electric tracks with a solo and some double-tracks; and the lead vocal was accompanied by a single harmony line during the choruses. There was a lot to like in there, but what worried me straight away was the muffled, 'spongy' quality of the acoustic guitar capture — like they'd left a dust-sheet over the mic! Given the importance of these tracks to both the rhythm and the mix's mid-range texture, I knew that the remix would stand or fall on whether I could salvage them, so there was no doubt in my mind where to start work.
An equaliser was my first port of call, filtering the low end at 75Hz and then shelving off another 3dB below 220Hz to tackle a combination of the instrument's sound-hole resonance and the mic's proximity-effect bass boost. Shooting for some mid-range definition with various localised peaking boosts, however, yielded only less appealing alternative timbres along a spectrum from 'boxy' to 'brittle'. Equalisation simply wasn't a powerful enough tonal tool for this task. What I needed was additional mid-range harmonics to play with; in other words, I needed distortion.
I fired up one of AcmeBarGig's freeware amp emulations (Tamla Head) as a test of concept, running it in a separate channel as a parallel process. At the outset, soloing revealed something akin to Lemmy wielding a chainsaw, but a quick twiddle of the emulation's dual Gain controls swiftly brought the drive into the realms of subtlety, whereupon I started flexing the virtual amp's other controls at random in search of some mid-range emphasis. Once trial and error had furnished me with a suitably honky confection, I faded it back in with the undistorted signal to see how it combined. I concluded that it was a positive change, but a little more tweaking of amp controls and distortion-channel level was needed before I was ready to move on to the second acoustic guitar track.
To help fill out the combined acoustic-guitar texture, I deliberately chose to give each track a unique tone, rather than just duplicate the same distortion. This time I tried SimulAnalog's freeware Tube Screamer modelling plug-in, once more pulling down the gain to avoid overstating the harmonics, and upping the Tone control for a pokier mid-range. Mixing this in with the dry sound, however, didn't have as strong an effect as I was expecting — which is usually a clue that phase-cancellation is working against you. Sure enough, around 60 degree of phase rotation from Audiocation's Phase plug-in on the distortion channel made all the difference.
It's probably worth my making clear that the purpose of these distortions wasn't to obviously simulate 'amped' sounds, even though both parallel channels definitely sounded that way when soloed. Because these distorted signals were mixed in at low level, the effect was much more subtle than that, although still considerable in terms of changing the subjective timbral character of the instrument.
I followed Rob's lead by panning the acoustic guitars to opposite sides of the soundstage, but couldn't resist giving each a bit of additional width by locating the distortion return slightly farther out. While focusing on how the two parts interacted, I made further tweaks to the Tamla Head's EQ and return levels, but I still yearned for more sustain, so I inserted Focusrite's Scarlett Compressor onto both tracks, using its lowest 1.1:1 ratio with a low threshold and fast time constants, to give the whole dynamic range a transparent-sounding squeeze. That helped a certain amount, but I was only able to emphasise the low-level elements to my satisfaction via another instance of Scarlett Compressor in a parallel configuration, fed from both the undistorted and distorted tracks. This employed quite severe 4:1 compression, but I only mixed in a shade of it to avoid the sound becoming too unnatural.
I was finally out of the woods with the acoustic guitars, and any further necessary refinements could wait until I'd introduced some of the other parts. The hardened edge of the acoustic parts played into my hands when it came to adding the electric rhythm track, because it gave the timbres more of a logical relation to each other, somehow. In fact, I even ended up adding a little more distortion to the electric, this time from Aradaz's freeware Amp Crunch which I'm a bit of a fan of these days, just to give it a touch more presence. A couple of decibels of low mid-range EQ cut cleared some close-mic muddiness out of this track, but I deliberately avoided high-pass filtering the low end much, because I had an inkling that allowing some of that into the mix, as opposed to relying mainly on the bass down there, might bolster the subjective warmth of the whole production. It also allowed me to spread those 'warmth' frequencies further across the stereo picture, because there's less expectation that guitar parts should be panned centrally.
I wanted to keep the mix's middle ground clear for the vocals, but I wasn't keen to just shift the electric guitar off to one side, because I've never cared much for a lopsided stereo picture. So, instead, I opted to 'stereo-ise' the mono signal via a Haas delay setup: panning the guitar to the left and a tight echo of it to the right. This is a trick that needs to be used with caution, so let's consider some details. The first main concern is one of mono compatibility, because combining the panned signals to mono provides the ideal recipe for damaging phase-cancellation between them. (For full details about the mechanics of this, check out the in-depth explanatory article at /sos/apr08/articles/phasedemystified.htm.) Lengthening the delay time can reduce the severity of this, but the limitation is that above about 20ms the echo starts to be heard as a separate 'slapback' reflection, generating a potentially distracting left-right 'ping-pong' motion. In practice, therefore, Haas delays will always carry some tone-altering side-effects in mono, so it's wise to tinker a bit with the exact delay time to make this change as benign as possible. I tried a few different settings, while switching my monitoring back and forth to mono, eventually judging 18ms to be the sweet spot in this situation.
Wide panning of the source and echo will also exacerbate stereo/mono differences, so I suggest narrowing it for most practical applications. Bear in mind, too, that even if the source and echo are at the same level, the panning of the source will still tend to dominate the resultant stereo image, so if you want something subjectively 'centred', you have to pan the echo further off-centre. This was exactly what I did for this remix, moving the guitar part 30 percent to the left, but the delay 50 percent to the right. If the phase cancellation still bugs you in mono, you can further reduce its impact by high-pass filtering the echo, but I preferred not to do so in this case. Because I wanted to follow up on the idea of enveloping the listener with the low end, I was content to accept the comparatively small remaining mono incompatibility as a trade-off.
Another side-effect of Haas delays is that they cause the treated sound to recede slightly from the listener in the mix's depth perspective, but this worked in my favour, because I rather liked having the acoustic guitars more up-front to drive the rhythm. To a similar end, I also compressed the electric guitar quite hard — because it seemed to me that rounding off its level peaks a bit encouraged it to sit more comfortably behind the acoustics — and gave each of the acoustic guitars more upper spectrum using a touch of psychoacoustic enhancement from Reaper's Jesusonic Exciter plug-in and couple of 2-3dB boosts from ToneBoosters TB_Equalizer (not the same ones on each channel, mind you, because different settings seemed to balance the instruments better in stereo).
Now to see how the bass fitted in with what I'd done so far. The DI's raw sound was eminently usable, although curtailing its extended HF response did help prevent finger noise and string rattle from intruding unduly on the mix's foreground. ToneBooster's TB-Compressor supplied all the dynamics control I needed via 4-6dB of gain reduction at a 4:1 ratio, with fairly relaxed time constants of 40ms attack and 120ms release. The only real bone of contention was the 200-400Hz octave, where the bass and electric guitar were fighting each other, giving rise to a woolly low-mid build-up. How much of this region the bass instrument can afford to lay claim to is a key decision in any mix, so I tried out a few different balances for this, before deciding to cut a couple of decibels from both tracks.
With a much stronger backing texture in place, I figured I'd get the lead vocals in next so they wouldn't have to fight too hard for space. Proximity effect on the mic was muddying the low mid-range, but was handled easily enough with 130Hz high-pass filtering and another 6dB cut at 240Hz. The levels were also contained pretty effectively with a single Scarlett Compressor operating at 3:1, although, admittedly, I did drive it hard (10-15dB of gain reduction) to lend the delivery greater attitude, and this led to some sibilance problems. An instance of DDMF's NY Compressor with a 7.5kHz band-pass filter in its side-chain provided most of the cure, but for those esses that hit the top octave especially hard, I inserted Toneboosters's split-band TB_Deesser to target that region directly.
The biggest hurdle, though, was a harshness in the presence region which wearied the lugholes at any reasonable fader level. The difficulty was that it came and went, so although I was able to make some headway with a 2.5dB peaking cut at 4.5kHz, any more cut there dulled some syllables unacceptably. Even setting the dynamic EQ in 112dB's Redline Equalizer to compress the offending zone offered an incomplete solution, so I had to take up the slack by automating an EQ manually, dipping the presence region and/or the whole high end whenever a particularly piercing note came along.
To compensate for all the high-frequency cuts, I added a broad sheen of 18kHz shelving boost from Variety Of Sound's freeware Baxter EQ, which is rather good at that sort of thing. Despite this, though, the lead vocal was still sitting a fraction behind the guitars, so I set up one more parallel compression channel to address that: one of my old freeware favourites, Mokafix NoAmp. As usual, I backed off the drive and got busy with some additional EQ to keep its contribution restricted tastefully to the most useful high-frequency regions.
I now had a clear idea of how much space would be available to the drums, so I moseyed through the tracks available. The raw overheads were in good shape, and fit straight into the arrangement with only a 4dB dip at 10kHz to distance the cymbals. Even the kick's low end sounded pretty tight through these mics (albeit light on the lows), which is something of a rarity amongst Mix Rescue submissions, so I didn't need to apply any of the low-end cuts that are normally required to stave off LF flab.
The stereo image was too wide for me, but I noticed that the 'room' mic wasn't much more ambient than the overheads, so I left this unpanned and mixed it in as a 'centre fill' to narrow the image. The third mic did, however, have rather a woolly low end, and had caught a lot of out-of-tune snare resonance at 273Hz, so I high-pass filtered and notched this using ReaEQ. In addition, I experimented with some small delay offsets between the room and overhead mics, to see if that might provide a more solid combination between them. Around 24 samples of negative delay seemed to provide a marginal enhancement to the snare sound, so I left that in.
The difference between the kick sound in the overheads and the sound coming through its close mics couldn't have been more pronounced. The latter was pillowy and bloated, with several lingering low-end resonances — the most serious at 78Hz and 115Hz. I tried zapping some of these with EQ in an attempt to tighten up the sound, but I quickly came to the conclusion that I was throwing out the baby with the bathwater by perforating this critical frequency range. Abandoning that angle, I turned to gating instead to remove the unwanted sound between hits.
The setup involved pretty basic hard gating, so once I'd set up the threshold level to get the gate opening reliably, the main concern was the gain envelope's time constants. You normally want a fast attack for drums, to preserve initial transients, but if you've got a lot of spill, such that you need a high threshold setting, the danger is that the lower-level attack onsets of softer-edged drums will be unnaturally truncated — which was exactly what I encountered. Sometimes I welcome the additional hint of 'click' definition this brings, but it seemed out of place in this mix, so I engaged the gate's lookahead feature to restore a more representative attack profile.
I did still feel, though, that the sound would benefit from a much meatier initial 'thump', something I first tried to add using SPL's Transient Designer, but with limited success. In retrospect, I should have known it wasn't the right match, because I find it's always better at providing transient definition than weight. So I left in just the 3dB of attack gain that was actually beneficial, and continued my pursuit using a slow-attack setting in Scarlett Compressor. The trick to achieving more thud with compression is to pile on the gain reduction (I was almost end-stopping the 30dB gain-reduction meter with my 6:1 ratio!), while leaving at least 20ms of attack time so that the front end of each hit can break through before the compressor really reacts. The precise attack time you use can have a great deal of impact on the subjective bassiness of each hit's leading-edge pulse, so you need to listen carefully to the low end as you adjust that control.
The subjective length of the kick-drum hits was also an important issue for me, so I spent a few minutes adjusting the compressor's release time, in conjunction with the gate's Hold and Release controls, in search of the best compromise between appealing sustain and unwelcome spill/resonance break-through. Because of the low end I'd left in the overheads channels, I made sure to check for the best polarity combination when mixing in the processed close-mic, and it was good that I did, because the inverted setting clearly trounced the default in terms of LF power. Finally, I shaved away a couple of decibels at 100Hz to avoid too much overlap with the bass guitar.
Further lookahead gating and heavy compression proved to be on the menu for the snare, even after I'd high-pass filtered some unpleasant low mid-range mud and notched out unhelpful ringing at 274Hz and 532Hz. This wasn't some kind of 'cookie-cutter' approach, though, because there were several crucial differences between each channel's settings. The snare gate, for instance, was only set to a -6dB gain range, because I find that too much level-modulation of snare-mic spill (especially that from the hi-hat) typically does strange things to the kit's internal balance. The compressor, on the other hand, was set up with minimum attack and release times to generate maximum density and sustain during each gated burst — almost the opposite of the attack-boosting patch I used for the kick.
The tom mic had already been gated, and only needed a cut filter at 44Hz to keep the low-end under control and a broad 5dB cut at 515Hz to recess the mid-range, which was already coming through the overheads plentifully. With the full kit sound now in front of me, the only thing I remained doubtful about were the cymbal sounds, which came across as a bit cheap and ragged for some reason. I was reluctant to start processing the overheads too heavily, though, because I liked the way they were putting the snare across, so I triggered some supporting samples instead, reaching for that old faithful, SampleLab's Drum Fundamentals sample library.
I imported a couple of likely specimens and spent a few minutes adjusting the pitch and EQ of each to try to most effectively supplement the existing overheads presentation. If you plan on doing this on your own projects, I'd suggest not using a pitch-shifting algorithm for this, because it'll tend to make the noisy cymbal tails sound rather phasey and will almost certainly reduce high-frequency sparkle into the bargain. Better to reduce the playback speed of the sample, much as the old hardware samplers used to do — it'll still change the apparent size of the cymbal as you shift, but to my ears that's a lesser evil.
Once I had a reasonable sound for one hit on each cymbal, I set about copying the samples for each successive occurrence. I didn't obsess about their exact timing, because I didn't actually need the transient of the sample, only its sustain, and the former was easily removed by applying 50ms or so of fade-in on the audio regions themselves. I also slapped on a fair bit of compression, using a high ratio in conjunction with a medium-slow release to increase sustain. The final step was to find a suitable level for the samples, which was actually quite subtle in the grand scheme of things.
With the exception of the electric guitar's Haas delay and a single 'character' effect for the solo, there were actually very few effects on this session, because by the time I'd reached a full dry balance, it honestly sounded fine to me as it was. However, after further deliberation, I found myself speculating as to whether a light dusting of short ambience reverb might help make everything feel a bit more as though it was performed together, so I opened up Magnus Jonsson's Smartelectronix Ambience plug-in, to see if I could rustle up anything worthwhile.
The 'Papen Dark Rimba Room' preset seemed promising, after I'd made a fractional adjustment to the Room Size parameter, to produce a less coloured tone. (I've now noticed this with a number of reverb processors: even when you have the right sense of scale to the effect, small changes to the Size parameter often still make quite big differences to the virtual room's tonality.) When I applied the treatment to a few of the instruments, it still sounded rather like the artificial add-on it was, but once I'd cut the decay time by a third and reduced the diffusion, it became much less audible in its own right. A 270Hz high-pass filter and a couple of decibels of HF cut to the return channel with ReaEQ completed the setup, and then I just had to ensure I wasn't adding too much by giving the return channel's Mute button a good workout, particularly towards the end of the mixing process.
And that was it, give or take a few tuning/timing edits; a sprinkling of Harmonizer-style stereo widening and tempo-echo on the lead vocal; a few decibels of squeeze from Variety Of Sound's Density MkII on the master bus; and a healthy helping of fader automation to refine the instrument levels from moment to moment. Referencing the mix against commercial releases also showed that some overall 'smile curve' EQ would be sensible — I'd been acclimatising to a new monitoring system, which had led to my instinctively undercooking the frequency extremes. A first draft I sent to the band also elicited a request for a bit more mid-range cut-though to gently highlight the part's melodic content, but a couple of decibels boost at 1.2kHz sorted that in two ticks.
As this month's remix demonstrates, there are few better processes for enacting tonal changes than distortion, so don't assume that EQ is the be-all and end-all in that regard. Hopefully, I've also been able to show that processing and effects techniques can't be learnt by rote, because there are times when 'standard' trends can justifiably be turned on their heads — as in the case of this remix's distortion processors outnumbering its reverb/delay effects, despite the primarily acoustic genre.
The Wrong'uns are Ricky Contreras (drums), Marc Helliwell (bass), Steve Robinson (guitars), Rob James (guitars), and Adèle Johnson (vocals). Formed in 2009, they initially performed covers, but gradually made the transition to playing original material. The song 'Rothko' was written by Steve Robinson and recorded at a local studio in Wokingham, and brings a grittier electric element to the band's previously more acoustic sound.
Steve Robinson: "The first reaction from the whole band was 'Wow!', followed by 'How much will it cost to get all our tracks remixed by Mike?' Overall, the new mix gives the song an atmospheric texture that elevates it beyond the generic sound of the previous versions. The space between the acoustic and electric guitars now opens up the sound, and the vocal lyrics come across clearly.”
Rob James: "You can hear the nuances of the acoustic guitars and bass, and there's now that hint of crunch which was lost in the original mix. The vocals are now much sweeter, and the backing vocals add spice to the chorus without overpowering.”
Ricky Contreras: "The bass drum in the remix now sits much better and is no longer overpowered by the bass guitar. The other drums and cymbals are more crisp and clear, and are generally at a better level.”
Adèle Johnson: "The vocals are a vast improvement, now floating on top of the instrumental, rather than having that 'singing in a cave' effect, as in the original mix. What I love about the remix is being able to hear and appreciate every separate element, despite everything gelling together so much better as a whole.”