Mix Rescue: I Strip For Couples

Neil Rogers at the mixing console, while tracking bass for this track.

Mixing strings in rock music: our engineer follows up last month's string‑recording session by describing how he integrated those strings into a full‑on rock mix.

In last month's Session Notes column (http://sosm.ag/apr14-sessionnotes), I described how I approached the recording of a small string section for Jay Williams' project, I Strip For Couples. Rather than 'rescue' a track this month, I thought it would be useful to look at how I went on to process what we captured during that session, and how I got the strings to sit and work effectively in the full mix, alongside drums, guitars and hip‑hop vocals — because I know several readers who've struggled to weave strings convincingly into such tracks.

To re‑cap very briefly, Jason wanted big-sounding real string parts for his project. There was no real budget to speak of and so, although we had a great room to record in, we only had a string section of six players with which tomake a sound that resembled something like an orchestra! This was achieved by working out three layers, or takes, that combined musically to create the effect of a bigger production. Whilst I think we did pretty well in achieving that objective, finding room for these big strings — alongside drums, bass, guitars, keyboards and vocals — was to be no mean feat.

Working Out A System

For a project like this you don't stand a chance of getting any level of control over proceedings without doing some organisational legwork first. The organisation of the drums, bass and guitar elements of the track was pretty standard stuff, which involved stacking things in my preferred order in the arrangement window and creating clearly labelled groups. For the strings, however, I wanted to be able to organise them in such a way that I could experiment with a few different techniques and combinations of takes. It was important that I'd be able to do this without having to every few minutes to work out what was what.

The string section consisted of three violins, one viola and two cellos, so when the three layers (takes) were combined, that gave me the equivalent of an 18-piece string section to play with in the mix. Each player had been close-miked and we also had two stereo room options, giving me a total of 32 tracks to organise in a meaningful way. At this stage, my mind was full of possible routes I could take to enhance the illusion of a more traditional orchestral setup, and some of them were elaborate, but I reminded myself that this was all pointless without hearing things in the proper context, so I got on with my temporary task of session admin in my Ableton Live DAW software.

First I made a quick sketch on paper to remind myself where all the players were positioned in the room. That would help me to find an initial position for the close-mic tracks in relation to the room mics. I then went through, re‑naming everything and implementing a simple colour-coding system, before assigning the three takes of strings, complete with the close and room mics, to their own groups. I created three groups, one for each layer, each of which contained separate submix channels for the close-miked sources and the room-mic options. Once I'd done some basic panning and balancing of the close mics, I was left with a mere six faders that would afford me an easy way to audition different balance and placement options when mixing, but also with a quick and clear system that would allow me to dive back in and adjust individual elements should that prove necessary.Working out a meaningful, colour-coded system for organising the 32 tracks of strings was essential.

Building A Rough Mix

Before exploring any specific processing, my first thoughts with a mix are to see how far I can take things using just the faders, creating and a good, old-fashioned static balance. Yet, for this particular track — 'FaultJam (1974)' — the strings don't enter the fray until about half‑way through, so I felt it was important to set the scene a little before deciding how the strings would be accommodated. I knew that so much of this production was going to be about managing space and headroom in the mix, and due to the style of the track, the key elements of drums and bass and vocals had to have a good amount of size, power and credibility to them, or the whole thing just wouldn't work. As I'd had the pleasure of recording the drums myself and had worked with drummer Mick Roman before, I had no excuses!

Drums & Bass

Mick is a joy to record. He hits the drums hard and very cleanly, and the only challenge with him is that he likes to really lay into his crash/ride cymbals. That can sound extremely good when he plays live, but it can often create a few problems when mixing, due to the level of cymbal bleed on all the close mics. You can get a drummer to hit a little softer, of course, but I also find that the slightly darker overhead sound you get from ribbon mics can be helpful, and this, along with the typical close-mic setup I'd used, allowed me to create a good drum sound in no time at all. Mick had chosen to use two snares, with a clear change in style part way through the track, so I tried to organise and group things with this in mind.

In this mix, I chose to focus on the mono mic that I'd positioned in the corridor outside the recording room, as I felt this had plenty of vibe. Because much of the sound had to pass through something physical to get to this mic, it has the effect of naturally filtering off much of the top end — and that meant I could compress the living daylights out of it before the cymbals started to wash everything out. After a little experimentation I ended up using Sound Toys' lively Devil Loc Audio Level Destroyer plug-in, which created a wonderful sucking, rhythmic compression/distortion.

This on its own sounded a bit wild, but when blended in with the regular kit elements it provided an instant and adjustable amount of attitude. Despite my best efforts with phase manipulation and some EQ, though, it did mean that I lost a little low end. To counter this, I turned to the excellent ReFuse software Lowender plug-in, augmenting the kick-drum channel with just a whisker of 30‑50Hz, followed by a 25Hz high-pass filter to catch any wayward low frequencies. This resulted in a full-sounding kick that might suited the hip-hop-style vocals. The phase of the drum channels was checked against the overheads, while ReFuse Software's Lowender plug‑in slightly exaggerated the kick's natural low frequencies.

The bass guitar enters the track at a similar point to the strings, so I started out by doing some quick processing and balancing it nicely against the drums. I had both a clean DI and a separate DI via a slightly driven Tech 21 SansAmp pedal to play with. You wouldn't think there'd be much phase difference between these two sources, but there was. I performed my usual check of polarity-inverting one signal and then using a phase-rotation plug-in (Betabugs Phasebug) to find the point of greatest cancellation. The beauty of this method is that you can then simply re‑flip the polarity to get the fullest sound possible. Doing this revealed a surprising and pleasing amount of extra low end and general solidity. I swiftly applied a 'safety' high-pass filter at around 25Hz and a few dB of smoothing compression before turning my attention to the vocals.A little phase rotation, courtesy of Betabugs' freebie Phasebug plug‑in, revealed a surprising amount of extra bottom end on the bass guitar.

Vocals

I find that when working with hip-hop or rap vocals, it's best to keep the processing quite simple unless you're shooting for a particular effect or aesthetic. Again, I had the pleasure of recording a good artist, Lavern Ashley Rhodes, myself, so I'd already selected what I thought was the right mic and recording chain. I'd also applied some compression at the recording stage, as I felt confident it was working well and, if I remember rightly, I used a fast attack and medium release setting, which imparted a nice sense of urgency.

We'd recorded a good lead vocal and the obligatory double of certain words on the first or last beat of the bar, but I felt quite strongly that we shouldn't use those; it all seemed a bit clichéd to my ears, and I liked the idea of a single voice telling the story. I went with my instincts, thinking that I'd see how Jason felt about it when the mix was more developed. I applied a high-pass filter at around 100Hz, adjusting the frequency by ear to find the right spot, and applied a touch of de‑essing at 4.6kHz, just to tame a few bits of harshness. With the tiniest amount of short plate reverb, I then balanced the vocals against the drums and bass. The final EQ settings used on LaVern Ashley Rhodes' lead vocal: after an initial high‑pass filter, a slight boost at 1‑2kHz helped with intelligibility.

Working In The Strings

With things shaping up well, I felt it was now time to bring in the strings. As well as my initial organising of the strings, I did some basic high-pass filtering to remove any unnecessary low-frequency information, with the various filters turning over somewhere in the 30‑60Hz region. I also did some basic polarity checks, just to get a feel for if there were any gains to be had, but these showed very little difference.

Brief experimentation told me that there would be little gain to be had from mixing and matching the room mics, so I settled on the Blumlein stereo ribbon-mic configuration for all layers. I was keen to try and keep as much of this room-mic sound as I could, as it sounded so much smoother and nicer than the close mics, with none of their harshness. Also, the room mics had done a good job of capturing what was a great-sounding room. I panned the close mics in broadly to match the natural stereo spread, opting for just a tiny bit of extra width, and I also took the opportunity to nudge down the level of the odd player who wasn't quite nailing the performance.

It was interesting at this point to see how dramatic the smoothing effect was of having 20‑30 string voices playing at the same time. Listening to some of the soloed close mics was not pretty but as a whole it seemed to work fine. This 'realness' is one of the hardest things to achieve with string sample libraries, and why they often sound 'too good'.

With a quick initial balance of the three layers of strings now in place, I began to bring the strings as a whole in against the other elements in the mix, to see how things were sitting together. Immediately, I was aware that I was facing a slight tuning issue, in that although they sounded fine against themselves, at the business end of a mix there were a few 'sour' moments, as I like to call them. There's nothing like a slight tuning issue to play with your mind in a mix: we can often find ourselves battling against our own ears, which naturally like to re‑adjust themselves to a sound we might, at first, have found unpleasant.

I thought a possible solution might be to try and identify some of the worst offenders and try out some tuning processors, but although I was able to find a few likely candidates within the close mics, both the artifacts from the spill and the fact that I'd have to significantly lower the room mics meant that this option quickly became a non-starter. I decided, therefore, to put my faith in the faders again and, after allowing my ears to re-calibrate during a lunch break, I spent a while experimenting with different balance options within the three layers of strings. After a while, I thought I'd managed to improve how things were sitting. How I got there surprised me, because I'd initially thought the violins were the issue, but it was in fact the lowering of the cello close mics that made the most difference. On reflection, that's perhaps less surprising, given the frequency overlap between the cello and the bass guitar.Lowering the balance of the cellos seemed to help to improve comparative tuning of the strings.

Working The Space

I wondered if I could further enhance the illusion of a bigger string section by adding an extra (ie. artificial) sense of front-to-back depth. With this in mind, I set up three reverb sends, each one with progressively longer reverb characteristics, and experimented with trying to position the different layers with different amounts of these reverbs, to see if it felt more or less 'orchestral'.

This was clearly one of those ideas that may have been good in theory, but in practice it just made things seem more artificial, as it removed the natural sense of unity between the takes and spoilt much of the natural goodness provided by the room mics. That lesson quickly learned, I moved on.

Bearing in mind that a good proportion of my drum sound was mono, due to the level of the room/ambience mic, and that the other main components were bass and vocals, it's not surprising that the biggest area of congestion was in the middle of my mix. One good thing about this, though, was that there was a natural answer to the question about where the strings should sit, so I began to focus my efforts on finding a way of clearing some space in the centre of the string section.

It was easy enough to manoeuvre the close mics away from the middle but the stereo room mics needed help. I find Waves' Center plug-in, which gives you control over a stereo signal's M/S balance, can be quite useful in this scenario, although it's easy to overdo things. Using its simple controls, I was able to pull down the central level of the stereo room mics. I used another instance of the plug‑in on the main reverb send, again to give me a subtle sense of extra width.Waves' Center plug‑in was used to bring down the mid-level of both the strings room mics and the string reverb. This worked really well, and after checking in mono and carefully trying not to upset the balance I'd achieved earlier, I felt like I was finally getting somewhere.

The strings had no compression at all on them at this point, and I found that gentle smoothing compression with about 3-4dB of gain reduction seemed to help them sit even better in the mix, and this would also allow me to control the dynamics a bit more with level automation. An aesthetic I was also keen to introduce, which I like on a lot of older string recordings, is the sound of subtle analogue tape saturation, almost as if you can hear the meters starting to bend a little. After auditioning a few plug‑ins, I settled on Slate's excellent VTM plug‑in, driven quite hard at the input, to help me achieve this. Final processing of the strings included smoothing compression and analogue tape emulation for a subtle saturation effect.

The Big Picture

At this point in the process, where things feel like they're starting to gel, I find it useful to take another short ear calibration break, before resuming at a sensibly low monitoring level, and trying to get a feel for the overall dynamics of the track. The arrangement for this particular track was quite straightforward. It needed to come out all guns blazing to a certain extent, to create a nice moment when the strings entered proceedings, but also to have somewhere left to go in terms of energy and dynamics when it came to the outro.

I know some mix engineers suggest working backwards from the 'biggest' section of the track, but although I follow the logic of that approach, it has never worked particularly well for me personally; it can just feels too unmusical. I also enjoy the challenge of 'finding' some extra juice if I really need to! I found in this mix, for example, that just riding up the level of a parallel compression track was enough to give the drums and bass an extra gear for the final section.

Final Touches

The vocal was sitting quite happily without a huge amount of manipulation and, quite surprisingly, needed only the tiniest touch of automation to make sure it held its place as the track got busier. I also added a slight delay, which I rode up at certain points to add a touch of interest, but this was all pretty subtle.

In my battle to accommodate the strings, I found that it created a nice effect by having the volume rising ever so slightly as they came in, but then really jumping up for the second half of the section. This sort of broad-brush automation can be so effective in a mix, but it was only possible because of the ground work I'd put in earlier to get the strings to sit properly.String automation was only possible because of the careful groundwork I'd laid at the recording stage!

It's worth repeating the importance of sensible monitoring volumes at the final stages of a mix. If things are going well, you're hopefully just left with the task of considering the final balance decisions, which for me nearly always involve the vocals or bass. I find this usually means a quick check on headphones, a few checks in mono and a surprising amount of time on my little Panasonic compact stereo — a 'grotbox', which I've positioned away from my monitoring position. I find that this gives a great change of perspective.

Conclusions

Other than the M/S manipulation to lower the level of the centre of the stereo room mics, the vast majority of the work to get the strings to sit in the mix involved plain-old balancing of levels, and the process I went though to achieve this would have been difficult without my initial work to organise the project. It's common to find that you have too many elements in a mix that are demanding to be 'big' at the same time or are occupying the same area, and there are a number of processing options available to help you massage and sculpt things into shape. In my experience, however it's well worth exhausting the power of a good balance before exploring such options. Often, the solution is as easy as taking a break!

I was really happy that I was able to keep the room mics at a healthy level in the mix and retain the natural sense of space that they were recorded in. My feeling is that we just about got away with the tuning issues, though I did take a little convincing by Jay and other band members that is was OK. At the end of the day, if they were happy then so was I. We'd discussed the option of blending in some of the sampled strings that the track was written with, and I think that we might end up having to go down that route for some tracks in the project where the strings are a bit more exposed. But it works well without them on this one.

Incorporating large live elements that would traditionally have been done in a large studio is ambitious when there's no proper budget for your production. Certainly, if money was not an issue you could employ musicians, engineers and producers well versed and drilled in the potential issues involved with larger string recordings, and getting them to work in a more contemporary project.

However, it's great to get out and experience the challenges of sessions like these, and I'd encourage engineers who work mainly with traditional band-type setups to find such ways of stretching and developing their skills.