Mix Rescue: 'Libre' by French rockers Neopren

The bass, like many instruments in this mix, was multed to two separate tracks, to allow different processing to be performed on the parts in the verse and the chorus/outro sections.

Our engineer reworks a busy mix, to make the chorus and outro rock like they should.

When I first listened to this month’s track, ‘Libre’ by French rockers Neopren, I heard a diamond in the rough. The four musicians had recorded their contributions to this jam-packed track remotely, and three of them then did their own mix. Although the band preferred different aspects of each mix, they weren’t entirely happy with any of them, and so invited us to help them out.

There were plenty of reasons for me to believe I could achieve a good result: the song was nicely written and arranged, and most performances were good, including the lead vocal and guitar solo. On the down side, a few unwanted noises and timing issues would require attention, and some sources sounded a little thin to me. But there was a more significant problem: the densley populated choruses just didn’t deliver their payload, instead sounding small and congested in comparison with the verses, which had more room to breathe.

What’s more, while the tempo of the song never actually changes, a few factors conspired to create the illusion that it drops in the chorus. The bass part drops in pitch, and is also far less busy than in the verse, with the notes being both fewer and longer-held. Similarly, the rhythmic contribution of the acoustic guitar in the verse gives way in the chorus to long, distorted electric-guitar chords, and a laid-back wah guitar. The latter sounded cool enough, but dragged against the groove. And the kick and snare lost their definition at a time when the strong hi-hat rhythm was succeeded by a wash of cymbals. Taken together this all robbed the choruses of vitality, so my biggest job would be to put that right.

Mix Prep & Initial Balance

I whiled away a good few hours setting up this mix session, but it was time well spent. Having imported the multitrack audio into Cubase 8 Pro, I moved bunches of related tracks together, set up my basic routing and configured a few starting-point send effects that I like to have available. These included a couple of simple reverbs: a short, mainly early-reflections patch in Cubase’s REVelation and a longer patch, based on the trusty ‘LA Studio’ preset in Cubase’s REVerence convolution reverb. Another instance of REVerence, running a large-stage IR, gave me a dedicated snare reverb, which I routed to the drum bus. I also rigged up a few parallel chains, including a compressor for the kick and snare, and separate distortion processors for the lead vocal, the snare and the bass guitar. Many plug-ins offer wet/dry blend controls, of course, but I prefer to use dedicated channels for such processes, as it makes it easier to filter and shape the result without screwing up the dry source sound.

Some of the processing used to reshape the bass in the chorus: pitch correction to counter the long notes’ natural pitch drift; a bass amp emulation; Sound Radix Surfer EQ; and a filtered two-stage parallel distortion effect.Having colour-coded the tracks and matched Cubase’s tempo to the song, I trimmed silence from the audio files, more to make it easier to see the arrangement and navigate the project than to reduce the strain on my MacBook Pro’s drive.

The next stage of mix prep for me is to build a rough faders-and-pan mix, all the while listening to the track and making mental notes about elements I might like to showcase or tuck out of the way. As usual, I started with basic ‘LCR’ panning decisions: vocals, bass, kick, snare, hi-hat and lead guitar (which didn’t play at the same time as the vocals) went in the centre, and other parts were panned left or right, in a way that made the stereo image feel balanced. The one exception to the hard-panning (though I’d do other tweaks later) was the stereo drum bus, which I set about halfway between the centre and the two extremes. Drums can sound lovely on their own when they’re made wide and spacious, but in the context of a busy rock mix in which a lot of other parts are placed out wide, wide-panned drums can sound a bit odd.

I decided I’d experiment with Klangfreund’s LUFS Meter, which I’d recently been sent for evaluation. This not only measures perceived loudness but can perform a gain offset to ‘normalise’ the source to a user-defined loudness. My plan (inspired by Eddie Bazil’s article last year on mixing to pink noise: http://sosm.ag/mixingtonoise) was to set all sources to the same loudness, while keeping plenty of headroom available on the stereo master bus. I was pleasantly surprised by the results — I could hear each and every source. While it was nothing approaching a ‘mix’, the rough balance was in the right ball park, so I could see this technique being a useful time-saver.

My final step of mix preparation was to go systematically through each source, removing rumbles, taming esses, checking phase-coherency and stereo imaging and so on. In terms of noise, there was nothing too problematic that a few high-pass filters and scissor-tool edits couldn’t solve, other than an overzealous finger-squeak on the bass in the intro. EQ did a tolerable job of removal but the collateral damage was too great, so I used Steinberg WaveLab 8.5’s spectral editor. I don’t use such surgical measures often, and left similar noises later in the track well alone, but felt it was essential here, as the part was particularly exposed.

The overheads didn’t put the kick and snare quite in the centre of the stereo field, an opinion I confirmed using Melda’s free MStereoScope plug-in. I corrected this by setting the stereo OH channel’s panner to Stereo Combined Mode, and bringing one channel in just a touch further than the other. Of more concern was the sound of the overheads. As well as a general splashiness in the cymbals, I wasn’t fond of the way they conveyed the kick and snare — the sound lacked authority. So I decided to high-pass the overheads to use them mostly as cymbals mics, and that I’d build the rest of the kit sound from the close mics. I also quickly checked for the best polarity for each of the close-mic channels, and checked to see if time-aligning the overheads and close mics would provide any improvement (it didn’t here, but it’s always worth checking).

The acoustic guitar’s slightly lazy character wasn’t entirely unappealing, but it did benefit from timing correction via simple edits. The addition of a tambourine part then helped to firm up the rhythm.Finally, I ironed out a few timing wrinkles. There weren’t major issues with any of the performances, but I did feel that tweaking the acoustic guitar would pay dividends. It didn’t want quantising, as the groove needed to feel a little relaxed. Instead, I focused on making the drums, bass, keyboard parts and guitar align neatly on the first beat of each bar, using Cubase’s scissors and audio-warp tools, and tidied up the odd errant strum elsewhere in similar fashion.

Despite the improvement, it felt like something was missing. But rather than snip and warp the life out of the guitar part I tried a favourite trick: I added a stealthy tambourine part. Let me explain... The initial mixes had spread this guitar across the sound stage with two opposite-panned parts, but I tried this and didn’t feel it worked well — the guitar was everywhere but nowhere. Instead, I panned the guitar to the right, and panned a simple tambourine loop opposite. This balanced things nicely to my ears, creating a sense of width and helping the groove without sacrificing the slightly charming laziness of the guitar part.

I must have got away with the tambourine because the band didn’t remark on it when I sent them my first-draft mix to the band. However, they did query the acoustic guitar panning and although they didn’t actually request that I change it, their comment moved me to re-examine things. Eventually, I re-introduced the DI guitar part alongside the tambourine on the left, but aggressively high-pass filtered so it was mostly just clicking plectrum, leaving the warmth and tone on the right. On occasion, the DI part combines with the miked one to drag the image slightly inwards, but while this is painfully obvious when listening to the guitar parts solo’d, it’s barely noticeable in the full mix. Anyway, I preferred the changes, so I left it that way!

Bigger, Badder Bass

Now to start mixing in earnest. My first job was to shift the energy balance in favour of the choruses, and to do that I needed to do something about the relationship between the bass sounds in the verses and choruses. I multed the bass to two tracks, one for the verse and the other for the chorus/outro. The essence of my plan was to distort the bejeezus out of the chorus part, to give it some much-needed ground-shaking attitude (I hoped it could sound somewhat synth-like) and thereby counteract the dragging effect of those sparser, longer-held notes. First, I needed to nail down the part’s pitching. Bass notes which are held for a long time have a tendency to drift in pitch, and while that can be a nice natural sound, in this case it jarred against the guitars and synths. Cubase’s Pitch Correct, a real-time pitch-correction plug-in, set fairly assertively, provided a quick and effective remedy.

My next task was to flesh out the bass tone, which I achieved first via processing on the chorus’ bass channel and then with parallel distortion. Cubase’s VST Bass Amp is a great-sounding plug-in and, after experimentation, I settled on its ‘Greyhound’ setting; I gather this is based on a Gallien-Krueger amp/cab combination, but its selection was a case of trial and error. One of the plug-in’s distortion pedals, placed before the amp, cooked the sound nicely and a post-amp chorus effect added width to the proceedings.

Needing to refine this basic tone with EQ, I turned to Sound Radix’s Surfer EQ. This clever plug-in tracks the pitch of the incoming signal and allows you to cut or boost that pitch’s harmonics — the curve moves along the spectrum as the incoming pitch changes. It’s brilliant for when you wish to massage a bass sound to shape without boosting or cutting specific notes more than others.

I’d set up the parallel distortion track at the outset, using Audio Damage’s neat freeware FuzzPlus3, preceded by a high-pass filter so that it would only add mid-range energy. After juggling its controls for a while, and refining the EQ settings, I felt I’d achieved most of what I wanted. That said, I’d return to this later and use a combination of Klanghelm’s MJUCjr compressor and Cubase’s Envelope Shaper to make the note onsets a little more prominent. A final touch was a compressor to duck the bass when the kick-drum sounded, just to help the kick cut through what had become a delightfully gargantuan bass sound!

With the bass now sounding more dramatic in the chorus, I needed to reshape its counterpart in the verse, to make the transition between sections more effective, and I was slightly less heavy-handed here. I used a little compression (4.5:1 ratio, slow attack and auto release) to yield about 3dB of gain reduction, before adding another VST Bass Amp instance (the Greyhound again, but with different mics on the cab, different settings on the amp, and no pre- or post-amp effects). This was followed this by an SSL EQ and Dynamics Channel Strip plug-in, just for a broad boost in the mid range and a touch more (low-ratio, low-threshold) compression. Again, I ducked the part to make way for the kick, this time going for more gain reduction — a decision that was due entirely to the different nature of the bass sounds in the verse and chorus.

The key bass effect in the verse was again parallel distortion, which this time added a nice spatial dimension. For this, I used two distortion plug-ins in series (on the parallel channel). First was Cubase’s Quadrafuzz v2 multiband saturation/distortion processor, with different types of saturation/distortion in different parts of the spectrum, and working in mono on the lower bands but with stereo enhancement on the upper ones. Audio Assault’s crazy freebie Defacer was the second. Both plug-ins take a while to get your head around, but they’re incredibly configurable and good-sounding toys!

Reworking The Drums

SSL’s Drumstrip’s HF Enhancer breathed new life into the cymbals sound in the overheads.Though the drum part had been played well, I felt the cymbal sound required major surgery. EQ brought some improvement but you can’t boost what’s not there, so I used the harmonic enhancer in SSL’s Drumstrip plug-in: I set the Drive and Amount parameters far higher than I’d eventually want them, swept the frequency knob to find the sweet spot, then I just backed off the Drive and Amount settings to taste. The result seemed perfect in the verse (though I’d fine-tune the Amount several times before the mix was complete) but a little too bright during the chorus. To counter this, I used a high-frequency shelf in Melda’s freeware MEqualizer, setting the shelf where it sounded best for the chorus (7.5dB of attenuation at a touch above 9kHz) and then automating the plug-in’s wet/dry parameter to transition in a natural-sounding way between the verse and chorus. (That way, if I wanted to make further EQ tweaks to distinguish the overheads in the verse/chorus, it wouldn’t require more automation).

To sort out the rest of the drums, I’d multed the kick and snare to different channels for the verse and chorus sections, and kept the hi-hat close mic for the choruses muted (using fader automation; automating the channel-mute sounded too abrupt), during which the hi-hat didn’t play — there was more than enough going on without the bleed on this channel to contend with.

Some processing used on the kick. A pair of compressors and a transient shaper combine to sculpt the envelope of the hits, and two EQ plug-ins perform complementary jobs.I needed the kick to sound tighter throughout. There are various different ways you can achieve this, but I find the best starting point is often a transient-shaping plug-in, and here I used Cubase’s Envelope Shaper to shorten each hit’s sustain. (After changing my mind several times throughout the mix, I arrived at a Release setting of -10.5). From there, I used EQ to tuck the sound in alongside the bass. Once again I reached for Melda’s MEqualizer and its wet/dry setting. I find it easier to home in on the frequencies I wish to boost/cut this way, and I can then blend between the unprocessed and EQ’d versions to achieve a balance that works. The result was a chunky (wide, almost 8dB) low-end boost around the 100Hz region, and an even bigger but narrower one around 4kHz, to emphasise the beater click. These combined with a narrow notch at 140Hz to tighten the sound a little, and to prevent clashes with the bass.

Boz Digital Labs’ freeware Bark Of Dog, a resonant filter modelled after Little Labs’ VOG, added further low-end heft, and a pair of 1176 emulations (Cubase’s Vintage Compressor MkII) in series completed the basic signal path. Each compressor knocked 2-4 dB off the kick, the first acting as a compressor (4:1 ratio, long attack, very fast release), and the second a limiter (20:1 ratio, fast attack, fast release). The limiter started out as a safety catch more than anything, to prevent the kick triggering threshold-dependent processors on buses further downstream, but when I tried bypassing it I missed its tonal contribution.

Snare Dressing

I used a whopping nine tracks in total to craft the snare sound, but there was a real logic behind the apparent madness! The biggest contribution came from a very simple parallel distortion effect, which was then filtered using Cubase’s in-built channel EQ.Can you believe that I used nine different channels for the snare sound? Neither could I when I prepared to write this mix up, but there was method behind the madness. Four tracks catered for the source audio: both top and bottom snare mics, each multed for verse and chorus. I then had one group channel (bus) for the combined snare sound in the chorus and another for the verse. The dedicated snare reverb and a parallel distortion took me to eight tracks and, finally, there was a parallel compressor shared by the kick and snare. All of this before the drum bus!

The last one, the parallel compressor, is the only one that made me think “did I really need that?” Yet it does do a useful job. This effect comprised a Softube dbx160 emulation followed by Cubase’s Envelope Shaper, with a boosted attack and shortened sustain. It gave me a single fader to govern the amount of ‘click’ from the kick and the snare, enabling me to automate the level to make the drums sound subtly different during the verse and the chorus, and to make the drums a little more prominent in the busier outro. Unlike EQ automation, this stuff is easily undone by muting the parallel channel or suspending its automation.

More conventional processing on the snare included EQ and compression. There’s little point telling you the precise settings: I set the EQ purely by ear and set the compressor attack fairly long to let the transients through, and the release short enough that the compressor fully recovered before the next snare hit. I again used SSL’s Drumstrip, this time to add both weight and brightness via its HF and LF Enhancers, and the snare parts in the verse and chorus were treated to different amounts of the same reverb sends — both the dedicated snare reverb, to give it a greater sense of scale in the chorus, and a little of the other two, just to help it sit in the same virtual space as everything else.

But the channel that contributed most to the final snare sound was (yes, you’ve guessed...) parallel distortion. The snare was getting rather lost in the mêlée at times, and I felt that more actual ‘snare’ in the sound would help it cut through. Alas, the under-snare mic didn’t really offer what was required, so I used Cubase’s bog modest Distortion plug-in to simulate it; it may not be the most esoteric of plug-ins but it can be wonderfully effective in this role. The trick, when using it as a parallel effect, is to follow it with an EQ to carve out any unwanted frequencies it generates. That meant rolling off everything below 400Hz, and engaging a broad-Q (1.0), big (8dB) dip at around 1.2kHz. I then sent different amounts of the verse and chorus snare channels to this distortion.

That leaves only the hi-hat sound and overall drum-bus processing to account for. Although I usually record a hi-hat close mic, I rarely use it when mixing — but I did here because it was important to the groove and didn’t sound so good in the overheads. An Envelope Shaper radically shortened the sustain, and seemed to suppress unwanted bleed. I also rolled off everything below about 600Hz and added an 11kHz bump, which I determined by ear was the area that made it combine best with the overheads. A final touch was to use ducking again, this time using Cubase’s multiband compressor triggered by the snare — the reason being that the hi-hat really didn’t need to be audible when the snare hit, and I wanted to preserve as much headroom as I could. In all honesty, I probably over-thought this, which is easy to do when you have an individual sound under the aural microscope. On the other hand, perhaps you should trust your instincts: if I bypass the effect, I find I like the snare sound less.

Slate’s VTM was used on the drum and master buses. It’s an excellent plug-in... but you might want to turn down the emulated noise, wow and flutter!The drum bus processing was simple, comprising an instance of Slate Digital’s Virtual Tape Machine, set to 30ips and with the unhelpfully high noise, wow and flutter set to zero. This smoothed things nicely, helping the snare in particular bed better with the overall kit sound. Later, I’d add a hint of low- and high-shelf boost on the channel EQ, as I worked to complete the mix jigsaw.

Acoustic Guitar

I’ve discussed the timing of the acoustic guitar already, but there were also some tiny pitching issues during the verses, probably due to how hard it was strummed rather than a failure to tune the instrument properly. I decided simply to disguise this by rolling off the frequencies below about 130Hz, so they didn’t clash so audibly with the bass and keyboards. Automating the level of the guitar then brought little ‘phrases’ in and out of focus, weaving them around others from the keyboards.

I also did a little more work to refine the general timbre of this guitar, as I sought to fit it in the emerging tonal aesthetic of my mix. Largely, this was a matter of using compression to control the dynamics, and saturation for tone. Again, I used two instances of Cubase’s Vintage Compressor MkII, one as a compressor (3-4 dB of gain reduction, 4:1 ratio, slowish attack, fast release) which helped things to ‘bounce’ with the drums, and another as a limiter (20:1, fast attack, auto release). Cubase’s Magneto 2 tape effect provided saturation; it’s perhaps not as authentic a recreation as the Slate and UA emulations I own, but it offers a pleasing warmth and thickness that they don’t, and it was just the ticket here.

Vocals

Given that the vocals needed to be the star of the show, I’ve waited an awfully long time to discuss them! But if I’d not put all that effort into laying the foundations of the mix, it would have been so much harder to get the best from the vocals. Also, I’d had them playing all the while, familiarising myself with them, and being careful to leave space for them.

There wasn’t a huge amount of level automation required for the lead vocal, but a lot of work went into automating the send levels to reverbs, delays and distortions.My main tactic for the lead vocal was to use EQ and parallel distortion (again... Klanghelm’s freeware IGVI this time) to bring out the character I wanted. It added a brightness and breathiness that appealed, as well as a sense of urgency. I also added a touch of low-end power, via SSL’s Vocalstrip, whose low-pass filter doesn’t just filter, but actually adds a slight bump around the turnover frequency. This sounded great but it unhelpfully emphasised the sibilance, which I tackled with the de-esser in Vocalstrip, plus an EQ notch at 11kHz via the channel EQ, and a more aggressive de-esser (mda’s free De-ess plug-in) on the long-reverb’s return.

Again, I used a pair of compressors in series, this time following Cubase’s Vintage Compressor MkII with Eareckon’s freeware FR-Limit 87. Finally, I added an instance of Vacuumsound’s ADT. Yet another freebie, this artificial double-tracking effect is one of my favourite tricks for thickening vocals. This channel, along with the main backing vocal and vocal-echo channels, was routed to a vocal bus, on which another Vintage Compressor knocked off only a couple of decibels to glue all the vocal parts together nicely, and which gave me a single fader to bring up the dry vocal sound should I need it.

Very little detailed level automation was required, as the performance, compressors and distortion took care of most of that between them, but I used lots of automation of the send levels to the various reverbs and delays. In this way, I worked to differentiate the sound in the verses and chorus, and to create spot effects on certain lead and backing vocal phrases, sometimes also cutting a few of the harsher consonants out of the delays so they didn’t distract from the dry sound.

Lead Guitar

A few different processors lent the lead guitar sound more attitude and brought it up front in the mix, while detailed level automation ensured it could always be heard without suffocating other details in the mix.I wanted to step the energy levels up a notch when the guitar solo came in. I loved the part but the tone was too ‘polite’ to provide the necessary impact. I had two tracks to play with, the amped version and a DI. I chose to use both and route them to a guitar-solo bus. I slapped another freeware plug-in (TSE Audio’s TSE808 Tube Screamer emulation) on to the amped part — once again the settings were all about experimentation — and another instance of ADT. These two plug-ins probably got me 75 percent of the way there but, still seeking more grit and interest, I ran the DI part into a compressor (just to control a little ‘spikiness’) and then NI’s Guitar Rig 5. I chose a preset based on a Vox AC30, turned off the delays and tweaked other parameters to taste. Another instance of Magneto and yet another of ADT gave me a lovely thick yet bright sound when the result was combined with the miked-amp track.

With that, I had the guitar solo’s tone in the bag. Now to really work the send effects, sending the amp and DI tracks to different amounts of the delay and reverb sends. None of this took particularly long, but I spent what felt like an age automating the level of the solo-guitar bus, identifying which little phrases and melodies I wanted to bring to the fore, and where it was slightly less distinct, tucking it away to let another instrument grab the attention. Of course, this approach meant that the delay levels weren’t affected by the overall level of the dry part, but I didn’t feel that was a problem: as the level rises the sound gets drier, almost as if the guitarist is taking a step towards you. Or maybe that’s my imagination — you can get a little too close to a mix — but I do like the result!

The Rest

The other guitars were trickier to accommodate. I decided the rhythm parts would play a less obvious role than in Neopren’s mixes, partly because the big, distorted bass provided that sense of distortion already, and partly because other elements gave the sense of height and width I wanted. So I robbed the rhythm parts of their bottom end and used them in mono, in the centre, to supplement the bass sound. I did opt for a little tonal tweakery, though, feeding them into an amp modeller (Steinberg’s VST Amp Rack), just to thicken them a bit, with only a touch of stereo width restored via the reverbs.

A wah effect in Cubase’s VST Amp Rack plug-in was automated to bring the original wah part under control.This all left me bags of space out wide for the wah guitar part and all those nice textural synth and string parts, and for the vocal delay returns to do their thing. I spent more time than I care to admit trying to nail the wah guitar part, as it always seemed either to dominate or disappear! After seemingly endless experimentation I reached the conclusion that compression alone wouldn’t suffice: the actual wah pedal movement wasn’t as I’d wished. So I used another instance of VST Amp Rack (a Marshall amp and cab model this time) and automated a virtual wah pedal in that, to weave the part in and out of the mix in a way that pleased my ears more. The rest of my efforts to put this in its place involved detailed automation of both the channel fader and send-effect levels.

The two ‘strings’ synth parts were automated to ramp up the tension.Work on the various stereo strings, keyboards and synths was fairly straightforward. It was largely a case of filtering, EQ bumps/cuts, and reverb sends, along with mostly broad-brush level automation. The one exception was the almost marimba-like Rhodes sound used in the chorus, for which I used a couple of bands of Cubase’s Multiband Envelope Shaper to emphasise some of the click and shorten the sustain, to make it feel a little more percussive.

Signing Off

This was a project I dipped into and out of in my spare time, and so it took quite a while to reach fruition. But I spent probably three and a half days in total working on this mix. That might strike you as an awfully long time, but sometimes that’s what it takes when you have a reshaping job to do. And something I’ve neglected to mention up until now contributed to the time it took me: I completed the whole mix using only a MacBook Pro, a decent converter, and a set of headphones (Sennheiser HD650s). For me, mixing on headphones is viable (particularly if you use Sonarworks’ Reference 3 headphone calibration system), but it always takes me much longer than when working in front of speakers in a studio, with lots more time spent checking meters, more time finding alternative playback systems to check it on, and so on.