If you're the musician and the engineer, it's easy to focus too closely on your own instrument, and not enough on the mix...
Most Mix Rescue candidates seek a glossy sound from an involved arrangement, but Judd Meng was experiencing the opposite problem — how to achieve clarity and fullness within a simple arrangement. Judd didn't want things over‑polished, and the mix had a rawness about it that was entirely in keeping with the intimate vocal, but he was also conscious that some intangible magic ingredient appeared to be missing!
Mix Rescue | June 2011 by Sound On Sound
The drums had been a particular source of aggravation. He'd tried several miking techniques, but the snare wouldn't cut through properly and the toms lacked fullness. The overall mix tone was also an issue, because Judd yearned for the warmth he could hear on his favourite Eddie Vedder, Muse, and Nirvana songs, but was unable to achieve that without losing sparkle and 'nearness'.
Listening to his original mix, I could tell that there was no 'silver bullet' solution, and that only a thorough overhaul of the mix could generate the kind of improvements that Judd was hoping for.
First Things First
What immediately struck me on looking at the track line‑up was that the most tracks had been given to the instrument that appeared, to me, to be the least important to the arrangement — and I suspect that Judd's 'drummer's-eye view' of the production meant that he'd mixed those tracks first as well. There's a limit to how hi‑fi you can make your drums when you want a single acoustic guitar and softly delivered lead vocal to take centre stage! I thought it made more sense to start work with the main guitar/vocal recording.
The good news was that the all‑important lead-vocal line had been picked up well. There was a little popping, and a slightly unpleasant low mid‑range resonance at moments, but a 90Hz high‑pass filter and a narrow 4dB peaking cut at 470Hz (both from Reaper's ReaEQ) swiftly cleared up those niggles. I wanted an obviously up-front, larger‑than‑life sound that would command the listener's attention, so I wasn't shy with the compression, chaining Universal Audio's LA2A emulation and Stillwell Audio's faster‑acting The Rocket to control the dynamic range and lend some obvious character. This created some problems with sibilance, but an instance of Digital Fishphones' freeware Spitfish solved those splendidly, albeit with a little automation assistance on my part.
The acoustic guitar presented greater challenges. It sounded as though the mic had been positioned somewhere around the 12th fret, so the recording was mercifully free of sound‑hole boom, but the overall timbre felt rather too thin to carry a track like this. There was a good deal of pick noise and fret squeak during the verses, as well as prominent string rattles as the guitarist dug in more during the chorus sections. Given the changing nature of the part, I decided to 'mult' the verse and chorus sections to different tracks, and began by addressing the more challenging chorus sound.
The most pressing problem was the overbearing string rattle and pick noise. A little attack reduction using
SPL's Transient Designer helped the latter, but the former proved intractable, even when I isolated the high‑frequency region with multi‑band processing. Frustrated at the lack of progress, I took a quick break before returning with a renewed sense of perspective — and quickly realised that I could simply tone down the guitar's high frequencies in this mix, as that gave the lead vocal more room to shine. So where a complicated plug‑in chain had failed to achieve a satisfactory fix before, simple 4kHz low‑pass filtering was all I needed!
After the troubleshooting, some tonal improvements were achieved through layering a number of fairly subtle processes. I often like what tube and transformer distortion can do with thin‑sounding guitars, and although my first experiment with Silverspike's Ruby Tube drew a blank, my second choice, Bootsy's Tessla SE transformer simulator, provided a useful enhancement. The UAD Pultec EQ (another processor I regularly turn to where I want to add low-end warmth without murkiness) delivered the goods nicely too: a gentle combination of 300Hz peak and 2kHz dip gave another incremental improvement. A final step was to breathe some life back into the filtered high end. EQ would have been no help, because it would have counteracted the low‑pass filtering, so I turned to Reaper's Aphex‑style Jesusonic Exciter. I pulled the high‑pass frequency down to 600Hz and drove it fairly hard, to create lots of additional harmonics based on the original mid-range frequencies, but only mixed in a tiny bit.
Judging when to stop 'enhancing' can be a tricky call, but I didn't see any point in proceeding further without hearing the sound set against the bass and drums. The verse guitar sound would need to survive closer scrutiny during the more exposed verses, so I gave that more time. Again, some low‑end boost from the Pultec was useful (with a good dose of 5kHz shelving cut), but full‑band transient reduction couldn't keep the pick noise in check, so I split the guitar's frequency range into two bands at 1kHz, using Reaper's Jesusonic crossover plug‑ins, and processed each with its own instance of SPL's Transient Designer. Not only did this improve the pick‑noise reduction, but it allowed me to add fullness to the note tails using the Sustain controls.
Taming obtrusive fret squeaks is beyond the scope of set‑and‑forget plug‑ins, and the only effective method I know is to manually apply high‑frequency EQ cut only when it is required. You can do this with automation, but here I chose to chop out the offending fragments and apply separate static EQ processing to them using Reaper's per‑take effects facilities. In each case, I cut 12‑24dB of high end using a high‑frequency shelf. This may appear extreme, but it's amazing how much processing you can get away with where such minuscule snippets of audio are concerned.
With the UAD Fairchild compressor emulation providing tonal thickening and overall level control, I was fairly happy with how the timbre was standing up. Mixing the vocal part back in, the spill between the parts didn't seem to be causing problematic phase‑cancellation, and the vocal's hard compression proved beneficial, as it rode up the guitar spill between the vocal phrases, filling out the sound whenever the vocal wasn't the focus of attention. However, the vocal compression drew a little too much attention to some of the singer's movement and lip noises in the verses, so I applied more region‑specific high‑frequency EQ cut during the gaps between vocal phrases. I could also have dealt with this issue by simply gating the lead vocal, but this would have been throwing the baby out with the bathwater, because, on the whole, the vocal mic's spill was helping the guitar tone.
Bass In Context
The main problem I had with the bass part was that its low end was running riot! A bass should have low end, but it usually pays to keep it under control, and to balance the low octaves with useful information further up the frequency spectrum. To give the frequency balance a more useful tilt, I applied a 290Hz LF shelving cut with ReaEQ, but with its cut‑off 'corner' softened using the Bandwidth control, so that it gave a gradual roll‑off from 2kHz downwards. To give the low end more consistency, I turned, as I frequently do, to multi‑band compression, this time from Reaper's ReaXcomp, compressing the sub‑350Hz region at a ratio of 1.6:1 with a low threshold setting, to give 5‑7 dB of gain reduction, before making up the lost LF level with that band's gain control. For overall level control, I then used the full‑band ReaComp at a stiffer ratio of 5.5:1, but with a very soft knee, for firm but progressive gain‑reduction. This not only evened up some of the playing, but also reduced a considerable disparity in the recorded bass levels between the verses and chorus.
Switching between my full‑range nearfield monitors and an Avantone Mixcube served to further refine the bass settings. Firstly, it became clear that the bass tone still didn't have enough mid‑range density to bring out its melodic elements on smaller speakers, so I added in a bit of Universal Audio's Precision Hz bass enhancement. This kind of processing is very much 'suck it and see', in terms of the particular flavour of added harmonics you choose and how much you mix in, but as long as you keep the underlying aim of the processing in mind, it isn't tremendously hard work.
The second thing I noticed was an unwelcome frequency build‑up between the bass and guitar parts in the latter's bottom octave. Feeling that the guitars were more important than the bass, I toned down that region of the bass with a localised 4dB peaking cut at 170Hz, which really seemed to help to settle the balance.
With the vocals, guitars and bass all in the mix, I began to bring up the drum parts. The overheads clearly carried the most appealing presentation of the kit — but they also conflicted with what was already happening in the mix! The main problem was an issue of frequency masking, in both the vocal presence region and the lower mid-range. I was able to address these areas of concern using two 6dB peaking‑filter cuts, at 400Hz and 5kHz respectively, from Melda's freeware MEqualizer. Following this broad‑brush processing, it became apparent that pitched resonances were another problem, and I engaged super‑narrow EQ notches at 277Hz, 538Hz and 833Hz to rein those in. After another quick break, I was surprised at how much I already liked the drum sound, even though I was still only listening to the overheads — definitely a good sign!
A couple of dynamics issues remained, though. There was some rather spiky stick noise on the cymbals from time to time, which might not have worried me in the context of a louder rock number, but I felt that it drew too much attention to itself here, and pulled the drums out in front of the guitars in the mix. Another Melda plug‑in, MMultiBandTransient, worked a treat, allowing me to smooth out the overhead transients above 3kHz.
The snare-drum close-mic came nice and strongly through the overheads, and had something to offer in terms of helping the snare to cut through better, especially at the high end, now that the two Melda plug‑ins had softened this region of the overheads. Adding in the close mic and inverting its polarity gave a slightly more solid sound, and I tackled unwanted resonances with ReaEQ notches and a dose of negative sustain from SPL's Transient Designer.
Beyond these technical considerations, I was underwhelmed with the close‑miked snare sound. As is usual with home‑studio drum recordings, the track was fairly dull and unnatural‑sounding, and blended poorly with the kit. This isn't a concern if you want the snare sound right up front, but this song didn't feel like the right environment for that. Rather than reach for the EQ, which only adjusts frequencies that are already there, I added brightness using another instance of the Exciter.
For snare length and blend, I turned to Melda's freeware MReverb, starting with a 'Small Studio' patch. I wanted to emphasise the snare's high‑frequency sustain, but also to draw the snare into the kit's own space. Rolling off the reverb channel's low end below 235Hz gave the required treble‑rich tone, and a narrowing of the stereo field helped keep it within the image presented by the overheads. Slightly increasing the room size made the reverb fit more convincingly with the overheads, and a narrow 8dB EQ notch at 446Hz carved out an unpleasant low mid-range resonance.
The Beauty Of Spill
Overheads rarely have enough definition at the low end for modern music styles, so the kick drum was next to be added in. I'd left the low end of the overheads intact to retain an authentic low mid-range, so it was important to consider the phase relationship between the overhead kick spill and the close mic — and it turned out that delaying the close mic by 160 samples significantly filled out the drum's low end. Beyond that, the only processing was a very gentle low‑pass filter rolling off from around 1kHz, rounding off a clicky tone that seemed to suit more of an up-tempo rock track, rather than the introspective mood of this song.
This left only the toms to mix in — and that was pretty much all I did! Judd had struggled to reduce the spill, but had effectively been shooting himself in the foot, because the bleed from these four mics filled out the kit sound beautifully. I decided to balance the mics according to their spill contributions, rather than according to the levels of the infrequent tom fills, and the sorted out any tom imbalances using a few spots of automation.
Knowing When To Stop
I now had all the main elements of the mix on the go, and began to consider how the few additional overdubbed parts, as well as any global mix effects, might enhance the final result. The bare bones of this song already delivered the lyric in a strong and impassioned manner, though, and I felt that there was a real danger of 'gilding the lily' (ie. of arrangement frills or production trickery undermining the intensity of the song's message).
Fortunately, there was scant requirement for effects, because the spill was already acting to blend the tracks quite well. I did yearn for more sustain from the guitars, once all the instruments were in place, so created a separate parallel compression channel for the purpose, using a super‑fast instance of Stillwell Audio's The Rocket to hammer out any transient information. The only global effect I added was a fairly understated room reverb from 112dB's Redline Reverb plug‑in, just to give a sense of communal space, with the high and low frequencies reduced to remove clutter and avoid any obvious audibility. A simple stereo tempo-delay lent some warmth and sustain to the lead vocal and chorus guitar, and I also gave the vocal a bit of my usual pitch‑shifted delay stereo‑widener, to lend it more presence in the stereo field — although I filtered that effect return savagely at the low end to turn it into more of a high‑frequency enhancement than any kind of chorus‑like sound.
The only other 'effects' were bus processors, comprising an instance of URS Console Strip Pro (providing overall high‑frequency 'air' lift at 20kHz and roughly 4‑6dB of medium‑slow compression from an emulation of the Focusrite Red 3) and an instance of Universal Audio's new Studer A800 emulation running on my UAD2 card. I've done a lot of work on A800s in the past, and my initial instinct, given the musical material, was to go with the Ampex 456 tape model at 15ips, based on my familiarity with the real thing, but this seemed to overcook the mix, and I ended up going for a much cleaner setting: GP9 at 30ips. Between those two master plug‑ins, I was able to get a surprising amount of extra mix 'glue' without any sense that the production was selling out on the music.
As for the remaining overdubs Judd had provided, I was very circumspect, and only brought in a few vocal overdubs to aid in the arrangement right at the end. Whenever I tried adding anything more, it felt like it was diluting the impact of that crucial central vocal performance. In fact, you could probably get away with adding even less than I did — it's a very fine line to tread. All that remained to finish things up was a liberal sprinkling of automation, stabilising the balance from moment to moment, as well as between song sections.
Leave It Out!
Some mixes are as much about what you leave out as about what you put in, so it's important to ask yourself with each overdub you add whether you're really serving the goals of the music. However, even in sparse and raw‑sounding productions, there's still a place for corrective editing and mix sweeteners, just as long as they're sensitively applied.
Editing For Correction & Enhancement
I did a fair amount of editing work on this track, which was challenging because of the spill between the guitar and vocal mics. I had to carry out timing edits on both mics simultaneously, so that the phase relationship between them (and hence their combined tone) stayed constant. You'd be forgiven for thinking that spill would have rendered any lead‑vocal pitch-correction out of the question, but in practice I was able to adjust a lot of the tuning without the cure being worse than the disease. I created a pitch‑corrected version of the lead vocal in Celemony's Melodyne Editor, bounced it down as audio, and crossfaded back into the uncorrected version whenever pitch‑shifted guitar spill made the guitar sound like a fugitive from a Chicago album! I also attended to a number of other little nips and tucks: fixing the bass 'flub' at the end of the first chorus; disentangling a lumpy drum-fill before the third chorus; patching a few words in the final chorus with better‑enunciated snippets from earlier on; and cleaning up the final fade‑out. My final tweak was to generate a double‑track for the chorus guitar part by copying and pasting parts between the chorus sections. This was only really possible because the lead vocal spill on the guitar mic during the choruses was pretty low. I did this to give the guitar arrangement a subtle lift when the drums arrived, but I was also conscious that I wanted to avoid anything too obvious, so I deliberately matched the timing of the 'fake' double‑track quite closely with the main track, to avoid telltale flams, and resisted the urge to pan the two parts too wide.
Rescued This Month
This month's Mix Rescue candidates are the band Blue Lit Moon, whose song 'Dad's Glad' was sent in by SOS reader Judd Meng. Judd handles the drums and recording duties, Eric Johnson plays bass/guitars, and vocals are provided by Kyle Deters, and all three are in the US Air Force. The session was captured in Judd's garage — with the help of some DIY acoustic treatment and a very understanding wife!
Judd Meng: "Mike, you're a wizard — I love it, love it, love it! It's like eating gourmet food after making do with frozen pizza for three months! Everything has been cleaned up nicely, while leaving it 'intact' overall, which was something I just couldn't seem to get right. The remix is much easier to listen to because it's not so 'crispy', but it's also opened out and warmed up. The acoustic guitar sounds so much better and has more body. The vocals have the subtle aggressiveness I wanted, but without going over the top, and they're sitting where they're supposed to be. The drums and bass work together without getting in the way. Everything has a space that I struggled to achieve, and a better sense of timing too.
"Although I'd originally suggested Mike add more 'studio wizardry', with backwards vocal parts and other such nonsense, he's taken the song where it needed to go. Just because you can do something, it doesn't necessarily mean you should!”
We've placed a number of 'before and after' audio files on the SOS web site so that you can hear for yourself the changes Mike made on each source and the full mix.