Unfinished business: our engineer reworks a track he produced several years ago, and finds he'd left himself with plenty more to do!
This month, I want to take you through my recent mix of 'Strange Sensations', a track by the band Captain. I say 'recent' but I actually recorded the track as part of an album project I produced for the band way back in 2012. The album was mixed by legendary producer Steve Osborne (Elbow, Gregory Porter, New Order, Happy Mondays, U2...) but was only released last year, under some sad circumstances (see the 'Featured This Month' box). Along with a couple of other tracks from those sessions, 'Strange Sensations' didn't make its way onto the final album release, though, and the band decided they'd like to develop this material as 'extra content' for their fans — similar to what we might have thought of as B-sides a few years ago — as they work towards another new album early next year.
I'd enjoyed working on this song and had been a little disappointed when it didn't make the album, so when the band initially asked me if I could work on it again, I was obviously very happy to do so. But I was also a little worried about the prospect of getting the original Ableton Live project up and running. Several years had passed since I worked on the album, and I was already a couple of computers down the line. I'd changed DAWs in that time as well, and remembered that we'd used a number of software synths that I've not used for years, along with programmed drum parts and audio that had been recorded to several different locations! Thankfully, while the main Ableton session resembled the Swiss cheese I'd suspected it would, I'd had the foresight to archive the whole session for possible re-mixing at a later date. Not only had I consolidated all the audio parts, but I'd also bounced all the soft‑synth parts down to audio and, to my pleasant surprise, they were even nicely labelled, with a few helpful descriptive notes! This was a valuable reminder of the importance of archiving and future-proofing projects — I shudder to think how much more time and stress this job would have involved had I not had access to the all-audio archived project. It's all too easy to overlook this sort of thing when there are lots of other demands on your time (children, for example!).
Opening the project afresh gave me a new perspective on the song, and it didn't take long to understand why it hadn't made the cut. I still liked it, but it now seemed overly long. In particular, the synth parts seemed a bit passive, and we should definitely have been more ruthless in terms of the number of parts we'd left in. It's not hard for your focus on such things to soften when working on a project over a very long time — in fact, if you're heavily involved in a project as producer, that can be a very good reason for using another mix engineer. With so much distance from the song, though, it was obvious to me where the arrangement and instrumentation needed a little nip and tuck. So I resolved to make these decisions quickly, before over-familiarity began to cloud my judgement.
To do this, I felt the need first to 'present' the song properly, so I spent about 45 minutes in broad-strokes mode, building a rough balance and organising the parts into meaningful groups. I then determined to focus on three or four key areas, and really tried to discipline myself to address these before disappearing down any mixing rabbit holes.
The first of these areas was the overall length of the song. Despite the arrangement being quite traditional (an intro, a couple of verses and choruses, a middle eight and an outro) the song was over six minutes long. I can't recall precisely what our thinking was — we'd probably been planning a fade-out on the super-long outro. But I now decided a proper ending would be better, and that the song could comfortably lose a good 60 seconds of outro and still retain the feel of the song somehow 'drifting off' (which I think was the original idea). I also decided to axe half of the drum intro and to lose a few bars of the intro before the lead vocal came in.
I then made a conscious effort to remind myself of the things people really like about this band, and very close to the top of that list is the combination of girl/boy vocalists Rik Flynn and Clare Meckled-Szembek. Their performances here were excellent — listening to them in isolation, where the middle‑eight section leads into a breakdown, they sounded truly beautiful, with bags of feeling. Nonetheless, I also felt that there wasn't a moment in the song where they really took centre stage. I had an idea or two about things I could do later to try and remedy this, but now I was at least in a position to pay some serious attention to the actual mixing.
The instrumentation for this song started life on a laptop before moving into a full studio, where live drums, guitars and a few other things were added. Blending real and programmed drums isn't always easy, and I remember a great deal of debate and experimentation took place before we ended up where we did. We'd initially wanted the drums to 'rub' against the programmed parts to create a certain effect, but in the end I had to align them to the grid for it all to work. Mic-wise, I had a number of good sonic options for the live kit, but for the programmed drums I'd obviously thought it best at the time to bounce what we had down to a single stereo file. Drummer Reuben Humphries had played the drums to the electronic versions and had kept things simple and tasteful, and after a quick balance of levels all the percussive elements sat together nicely.
Having recorded the live drums myself, they were obviously broadly to my taste already — but it really is amazing just how much better your engineering sounds when you work with a great drummer like this one! I initially felt that the low end needed exaggerating a touch, but that was solved by experimenting with mic combinations and carefully checking phase relationships. The drums had been recorded in my old, smaller live room with a pair of Earthworks omni mics as overheads, and the bottom end was almost sub-like in its low-frequency content. So in this case, it was really just a case of fitting the other drums mics around the overheads — it's always easier if you can pick a part of the drum recording to build the rest around in this way. I can't stress enough how important it is to focus on this sort of thing before you jump in with any EQ when you're mixing live drums.
After checking the polarity of the kick and snare close mic‑signals, I blended these in beneath the overheads, just to add some weight and definition. I sometimes play with Pro Tools' time-alignment tool, to see if this does anything useful, and here I found that delaying the close-miked kick track by 112 samples helped them to add a little punch and clarity when mixed in with the overheads. At this stage, I really needed to control the booming low end and decide how to shape the kit in the context of the mix. I'll discuss fine-tuning the drums later, but to tame the overheads a gentle high-pass filter (around 70Hz) and multi-band compression focussed below 200Hz provided just enough control.
I had a decent rough balance of the whole song at this point, and the live drums were sitting nicely with their programmed friends. (Too nicely, in fact; I knew I'd want to add some more character to the sound later!) While I was itching to get stuck into the vocals, I first wanted a plan for which synth parts to use and which to ditch. There were four bass synth parts in total, and I wasn't convinced that these all worked together — two were providing really low, sub-like bass, and closer inspection confirmed my suspicion that one part was clashing musically with the others. In fact, it probably wasn't supposed to still be there!
A decision mix engineers often have to make is whether the kick drum or the bass guitar/synth will occupy the lowest frequency range in the mix. Musical training can obviously inform such decisions, but I tend to go with a combination of gut feel and what I think will make the song translate better on real-world playback systems. For example, I'm always asking myself the question: would this song still work if I couldn't hear that bass line properly? The bass parts were also quite busy, which can be a problem when it comes to really low frequencies. So I decided that I'd give the low end to the kick, leaving a bit of room for the drums to be a feature on any decent playback device. The bass line would be more about filling in the lower mid-range in the verses, with the sub-like bass only coming in to give the choruses and the outro a nice lift.
I'd reduced the bass parts to three tracks, and while the lowest one needed to remain in the centre of the image, two of them didn't need to convey much information below about 80Hz, so I decided they could be opposite-panned about 65 percent from the centre. I also used a little multiband processing, to give me greater control over the bass' low end (see box).
There were still plenty of decisions to be made about some other synth parts and how prominent to make the guitar, but it was high time I got cracking with the vocals. As I said, the vocal performances were great and the recordings sounded nice, but on the downside it all felt a bit too much like a blank canvass — a little more excitement needed injecting! My approach was to mute many of the synth and guitar parts, and work for a while with just the drums, bass parts and the vocals playing. I wouldn't always mix like this, but I felt strongly that if I could get these three elements to sound exciting on their own, the whole mix would be pretty close to where I wanted to be, and the rest should fall into place.
So... just how do you mix to make a vocal sound exciting? The first, obvious answer is to get everything right at the performance/recording stage, but purely from a mixing point of view, we're not exactly lacking in creative options these days. Ultimately, it's a mix engineer's taste and judgement that earn them their keep (there's a lot more to it, of course, both technically and in terms of understanding the artists' own taste and judgement), and when using effects to enhance a vocal, you nearly always end up with a very non-technical period of throwing stuff at the wall until something sticks — in other words, until your experiments deliver something you like. I'll touch shortly on what effects I settled on, but one thing I've worked at recently is crafting a particular tone on a lead vocal that I hear on mixes by some engineers I really admire. There are always exceptions, of course, but to my ears vocals are often more tucked into the music at the moment than they once were, yet still with a very clear and pronounced mid-range that allows for them to feel 'forward' and for all the lyrics to remain clearly audible. Hopefully, you'll have some idea of the kind of vocal sound I'm talking about, but it's perhaps a more challenging thing to get right than you'd imagine, since it's not achieved with EQ alone. Boosting frequencies anywhere in the 2-7 kHz region on a vocal can make things sound forward, but it can also very quickly make things sound a bit nasty! People talk about using certain compressors for their 'tone' rather than purely for gain reduction, and saturation and harmonic-distortion techniques can also go a long way to achieving this effect.
For this mix, I got close to the sound I wanted with a combination of EQ, compression and some very targeted use of saturation. As with the bass, I multed the lead vocal, and heavily processed one of the two tracks to make the mid-range really stick out. At the heart of this was FabFilter's Saturn plug-in, which allows you to focus its saturation/distortion processing on different frequency bands. This combined really well with the results of the very cool new Viridian plug-in from Acustica Audio (it's powered by their 'volterra kernel' engine, which effectively allows them to 'sample' the sound of external hardware rather than model it algorithmically). The latter plug-in seems to allow lashings of additive EQ before it starts to sound remotely bad, and a generous 6kHz boost did the trick. I then balanced this 'mid-range' version of the lead vocal against the more natural-sounding one. Both tracks were routed to a group bus, with which I automated the overall lead vocal level and applied some effects, the most important being Soundtoys Little MicroShift, which combined with a little reverb and delay to give the vocals a nice sense of 'spread'.
After a while playing with different combinations of panning, effects and balance, I was happy with how all the vocals were sitting. They needed minimal automation except for a little (1dB) level bump in the choruses. But then I entered into quite a lengthy stalemate with the drum sound...
I mentioned that it's not always easy to blend real and processed drums, and it took a lot of fiddling to arrive at the finished result. One thing that really helped was Oeksound's Spiff transient processor, which I used to soften the electronic drums — this not only carved out a sort of 'pocket' in which the top end of the real drums could sit without them needing to be too loud, but it also helpfully brought down the level of the electronic hi-hat. Another thing that got the two parts working more nicely together was some sample layering: a 707 electronic snare sample reinforced the acoustic snare, and an old LinnDrum kick sample beefed up the real kick. (I used these sample layers fairly subtly, though.)
This whole process was another case of trial and error and massaging things into shape, and when you get a bit stuck like this, I can't stress enough the importance of not soloing the tracks too often! In this busy, composite rhythm track, I found that listening to the drum sounds in isolation was almost entirely irrelevant to how things sounded when played together!
A few stern words with myself allowed me to move on from the drums, and I then spent some time building the synth parts and guitar sounds around the rhythm tracks and vocals. There's not a great deal to report about this part of the mix except that I was careful to try and manage the available mix real estate by using plenty of low‑ and high-pass filtering on the different synth layers. Apart from some panning decisions and a little reverb, it was just a case of old-fashioned balancing of levels throughout the song — and not being afraid to hit the mute button, which I did quite a bit in this mix!
Once everything in was place and I was happy with how everything was being presented, one crucial final thing played a big role in the overall aesthetic of the mix: how I made use of saturation and distortion effects. I make a point of mentioning this because I often question just what it is that these tools are actually contributing; I try to not just use them blindly. I've already said how I'd employed saturation on the lead vocals, and I also ended up with quite a generous amount on all the drums and percussion elements — which were going through their own subgroup — via SoundToys Decapitator, a long-standing favourite plug-in. Throughout the mix, there was processing somewhere that was doing some kind of subtle distortion, and I also had two different stages of saturation on the stereo mix bus! So what was it actually doing? Well, as well as adding some 'character', I felt that in this mix it was helpfully blurring the lines between the live elements and the programmed parts. It felt like the right thing to do. It's OK not to be shy with this sort of processing sometimes, but do try to do it for a reason. It's also worth mentioning that I generally tend to achieve better results by painting in thin layers — spreading such processing around the mix, or across multiple plug-ins, rather than getting one tool to do all the work.
Speaking of painting, I recently had a client in the studio who paints for a living. He was explaining that when he gets a bit stuck on a project, he'll put the canvas somewhere he can't see it and work on something else. When he comes back to it (usually first thing in the morning), he makes a note of the very first thing that comes into his head and works on that. Providing you don't have a tight deadline, it's straightforward to do something similar with your DAW projects, and I found myself doing something similar with this song.
This really wasn't one of those mixes that just came together over a few hours, and if felt at times more like a puzzle that needed solving. Yet it seemed clear to me on every fresh listen just how important it was to get those live and programmed drums to pull in the same direction, and to get that lead vocal to really lead — to carry the song along. These elements needed to be really strong because the synths and guitars, in particular, never really make any major statements. The arrangement changes were also important, and I was only able to make and refine these decisions by finding ways to ensure that I could remain objective. There was also more than one occasion where I just had to get over myself — to lose the fear of losing a part that I liked in isolation but which didn't fit into the overall picture (yes, that drum corridor mic really did sound cool, but it had to go!). Once I'd figured out the fundamentals, the rest was a case of 'sculpting' other parts into combinations that worked, and blurring some lines with carefully judged saturation. Six years after we'd started, I felt that I'd finally wrapped up this track, and it was a good feeling to receive a very warm response from the band for this mix of a long lost song.
It's a fairly common practice in commercial mixes to mult parts — that is, to create duplicates on which different processing can be applied. I often use a similar approach coupled with 'frequency bracketing' to set up multiband processing for bass parts in particular. This involves making multiple copies of the same track and using high- and low-pass filters to restrict each to a different frequency range. I might, for example, wish to saturate or distort the mid-range, while keeping the very low end clean, or to heavily restrict the part's dynamic range in just one area of the spectrum while allowing the rest to 'breathe'. While there are plenty of dedicated multiband plug-ins, they lack the flexibility of this method — the routing facilities and delay compensation and EQ plug-ins of modern DAWs make setting things up the work of seconds, and allow you to deploy whatever plug-ins you want on the different 'bands' you create in this way.
This technique, which I used on this track, can give you very precise control where you feel it's needed — and it is particularly useful for any genre that demands dense arrangements or which needs careful bass management (eg. dance, hip-hop or the more technical styles of metal).
It's all very well reading about how Neil approached this mix, but you can also hear the mix for yourself if you download the ZIP file of hi-res WAVs.
Download | 126 MB
London and Brighton-based alternative rock/indie band Captain enjoyed chart success in 2006 with their singles 'Broke' and 'Glorious'. In the same year, their Trevor Horn-produced debut album This Is Hazelville reached number 23 in the UK album chart. A second album followed in 2008 before an unplanned hiatus (due to their label EMI's acquisition by a private equity firm) saw the members pursue various side and solo projects.
Captain returned to the studio with producer Neil 'Bugs' Rogers in 2012 to record their third album, which was mixed by Steve Osborne. Sadly, the album's release was delayed for a number reasons, and when guitarist Mario Athanasiou lost his battle with cancer in early 2016 the band decided to release the album through PledgeMusic — and in doing so raised several thousand pounds for the cancer unit which treated Mario at Guy's and St Thomas' Hospital in London. The album, For Irini, is available now.