Mix Rescue: Adding Presence

We help transport listeners from a small studio to the Big Easy!

Robin Phillips is a very fine pianist and singer, who is well known on the London jazz scene. His home‑recorded piano skills are currently earning millions of streams for jazz‑house artist Berlioz, while his vocal chops lent some class to our video feature comparing versions of the AKG C414. And late last year, Robin was wrapping up a project that was particularly special to him: his first album of original songs for over 20 years. This latest endeavour features not only regular bandmates from a number of his current line‑ups, but also London’s Soul Sanctuary gospel choir, big‑name jazz players from both sides of the Atlantic, and a string quartet. The first single from the album is ‘Ode To NOLA’, his homage to New Orleans.

Robin normally mixes and masters everything himself, but this particular track wasn’t cooperating. Given its importance as the lead‑off single from the album, he called me to ask for some advice. One thing led to another, and soon I had a very neat and well‑organised multitrack to download!

Close Calls

Visits to historic studios such as Sun Studios and Muscle Shoals Sound Studios had convinced Robin that recording musicians in a room together was key to getting the vibe he wanted. To this end, he has made use of recording spaces including London’s MasterChord Studio and New Orleans’ Marigny Studio. ‘Ode To NOLA’, however, along with many other songs on the album, was tracked in his own home studio. This is a very well‑equipped affair, with lots of tasty mics and outboard feeding a Focusrite Red 8Pre interface, but the live room and control room are both pretty compact. Not a problem for recording solo instruments, but enough to introduce some challenges when you want to capture a band live including a full drum kit.

When you’re tracking live in the same room, drum spill onto the vocal mic is inevitable!

For the initial live session, Robin had sensibly banished bassist Louis Thorne to the control room in search of a spill‑free upright bass recording, while he himself sang and played Rhodes piano in the live room, with drummer Claire Brock on a full drum kit very close by. Electric guitarist Neil Cowlan and Pinstripe Suit brass players Stacey Dawson (sax) and Sam Sankey (trombone) added their parts later in the same studio, and the whole thing was topped off with a Hammond overdub by Robin courtesy of his Nord Electro 5D, amped through a Vox AC30 and captured by Coles 4038 and Shure SM57 mics.

Louis Thorne’s upright bass was tracked in the control room, using an AEA R84A ribbon mic.When we spoke, Robin identified a couple of things that were proving troublesome at the mix, and perhaps contributing to a sense that the track as a whole was lacking in presence. Both the Rhodes and the electric guitar had been recorded through wah pedals, and whilst both sounds were very cool in isolation, they were treading on each other’s toes. The upright bass sound was impressive on its own, but its powerful low end was proving hard to handle in context. Finally, Robin’s live vocal mic had inevitably picked up a lot of drum spill. The Electro‑Voice RE20 suited his voice very well, and has cleaner off‑axis sound than most dynamic mics, but because the spill was being bounced around in a small space before arriving off‑axis on the mic, it didn’t sound great. Pushing up the vocal fader thus compromised the otherwise well‑recorded drum sound, and adding compression or high‑frequency boost to the vocal exacerbated the problem.

Small Is Booty‑ful

A “lack of presence” in a mix usually means a recessed upper midrange, often paired with over‑abundant low mids. This is an ever‑present risk with material tracked in a small studio. To minimise the pickup of room sound and spill, you inevitably mic things quite close, with directional mics, and this brings the proximity effect into play. In this case, for example, the bass had been captured by an AEA R84A ribbon mic, and the figure‑8 pattern had introduced a lot of bass tip‑up. The brass overdubs were close‑miked with dynamic and ribbon mics, so the musicians could play in the room together and minimise spill, but likewise the sound was a bit too warm to cut through in a busy track.

There was, therefore, a sense in which the multitrack ‘wanted’ to sound soft. The close‑miked bass, brass and wah guitar were naturally rich in lows and low mids, while any attempt to push forward the midrange brought out the worst in the drum spill on the vocal track.

I used to think of source separation as a technology in search of an application, but having discovered that it can separate wanted audio from spill, I’m a convert.

Divide & Conquer

The obvious fix for the drum spill would have been to re‑record the vocal, and Robin has plenty of top‑end mics with which to do it. However, the live take had that all‑important vibe, which would have been hard to recapture later. Ditching the live vocal would also have meant discarding Robin’s meticulously recorded video footage from the original session, which formed an important part of his promotional plans.

Fortunately, help was at hand. I used to think of source separation as a technology in search of an application, but having discovered that it can separate wanted audio from spill, I’m a convert. I dropped Robin’s vocal track into Hit ’n Mix’s RipX, and a few minutes later, was rewarded with impressively clean vocal and ‘drum’ tracks. If you don’t change anything, these separated tracks recombine perfectly to recreate the original; the further you depart from this, for example by reducing the level of the spill track, or processing the vocal track, the more you risk artifacts being audible. In this case, the separation was good enough that I could lower the spill fader by 7 or 8 dB, and also EQ it to remove the peaky upper‑mid and muddy lower‑mid frequencies that were clouding the mix. I was also able to send to reverbs and delays just from the clean vocal track, meaning there was no risk of spill feeding into those effects.

Thanks to Hit ‘n Mix’s RipX, I was able to separate the drum spill from the wanted vocal. The spill track (top) couldn’t be muted altogether without introducing noticeable artifacts, but it could be reduced to a manageable level.

With the unwanted parts of the drum sound reduced to a manageable level, I started work on the actual drum tracks. These comprised a spaced pair of Calrec 1050 pencil mics as overheads, a beyerdynamic M88 on snare, inside and outside kick drum mics (Shure SM52a and Neumann U47 FET respectively) and a Shure SM81 on the hi‑hat. There being plenty of hi‑hat in the other mics, this essentially gave me four useful tracks to work with.

Claire’s playing was first‑rate, the kit properly tuned and the mic placements well chosen, so relatively little mix housekeeping was required to arrive at a solid basic sound. In this case, that meant time‑aligning everything, bussing the two kick mics together and using Sound Radix’s Drum Leveler to control their dynamics, notching out a ring on the snare mic with FabFilter’s Pro‑Q 3 and employing some basic EQ to tighten up the low end.

For this particular song, however, I felt the need to aim higher. ‘Ode To NOLA’ was supposed to be a rollicking, swaggering, tribute to one of the world’s great musical cities, and it demanded more than a solid basic sound to drive it along.

Making Room

There are many ways you can add excitement to a drum recording, most of which involve compression or saturation. However, the effect of compression on drums has a lot to do with the space in which they’re recorded. Compression can make the room sound suck and breathe in a way that we perceive as exciting, because it somewhat mimics the natural response of our ears to hearing very loud drumming. In this case, though, the drums had been close‑miked in a small space; there was no room mic, and if there had been, it probably wouldn’t have been useful.

Sending to a suitable reverb, returning it to the drum bus and then applying compression helped to add energy and life to the drum sound.

Consequently, I decided to fake it. I’m a big fan of the ‘LA Studio Drum Rooms’ presets in EastWest’s QL Spaces II convolution reverb: a selection of short reverb patches that enhance and blend with dry drums better than anything else I’ve heard. I set one of these up on a bus, fed it both from the snare track and the overheads, and routed it along with the drum tracks themselves to a single global drum bus, where I applied compression. And not just any compression: McDSP’s APB system includes a plug‑in called Chickenhead that brings instant attitude to almost anything. Sometimes it can be too much, but here, the combination of studio ambience and aggressive compression gave the drums the X factor they had been missing, and completely obscured any lingering trace of small room‑itis.

Bubbling Under

Heard in isolation, the upright bass sounded really good, thanks to Louis’ excellent playing. However, this proved to be one of those occasions where achieving a sound that worked in the mix meant making it sound worse in solo! The problem was that if I set the bass fader at about the right level to fill out the low end of the mix, it was only the low frequencies that were audible, and the energy of the playing didn’t come through.

I set out to rebalance the tone, using FabFilter’s Pro‑Q3 equaliser and Pro‑MB multiband to apply some fairly drastic attenuation below 250Hz and a boost further up the midrange. This helped to rediscover the vitality and bring through the woody quality that is so characteristic of the instrument. I was then able to make the double bass much louder in the mix without overloading the low end, which is what I’d wanted to achieve. But this brought to light another issue. The transient snap of string against fingerboard had been audible even at the lower level, and now it was actually louder than the snare. It was thoroughly distracting, and undermined the nice drum sound I’d just crafted, so I decided it had to go. I used a second instance of Pro‑MB, set up to duck the mid and high frequencies in response to a high‑frequency transient, along with Oeksound’s excellent Spiff transient shaper. The result was a double bass sound that is unnatural in isolation, but works much better in the mix.

Two instances of FabFilter’s Pro‑MB dynamic EQ at work on the upright bass. The instance on the left is dealing with the string snaps, briefly attenuating the midrange and high end in response to a high‑frequency trigger. The other is controlling the relative levels of the midrange and bass.

Fortunes Of Wah

An unsual aspect of the musical arrangement on ‘Ode To NOLA’ was that both the Rhodes piano and the electric guitar were played through wah pedals for most of the track. A wah pedal creates big resonant peaks which move up and down the frequency spectrum, and can be challenging to mix. One minute the instrument is pumping out a wave of mud at 300Hz; the next, it’s drilling holes in your speaker cones at 2kHz. Two wahs going simultaneously means double the fun.

Thankfully, both Robin and Neil had operated their respective squelch machines with restraint, but even so, some work was required to keep both at a consistent level in the mix and avoid cluttering up the low mids. I used SoundToys’ Filter Freak plug‑in to trim away unnecessary high and low end on the guitar, and Sound Radix’s Drum Leveler (which, despite the name, works well on all sorts of percussive and dynamic sources) to even out its dynamics. On the Rhodes piano, I did something similar with FabFilter’s Pro‑MB. By emphasising slightly different frequency ranges in each case I was able to make the two instruments work together rather than fight each other. Since both were played all the way through the song, it also seemed natural to hard‑pan them to opposite sides, and this instantly made for a nice wide mix.

The guitar solo in the closing section of the song was particularly difficult to sit in the track. With the fader at any fixed level, some notes jumped out whilst others were inaudible. Aggressive dynamic EQ made the tone more consistent, but even then, it was necessary to do extensive level automation to achieve a relatively even sound, a process which took several mix revisions to get right. In this situation, it’s often a good idea to cut out the solo section and place it on its own track, so it can be treated independently.

Unexpected Delays

Once disentangled from drum spill, Robin’s lead vocal required very little work. I didn’t want to slam it too hard with compression lest the artifacts of the source separation become too obvious, so instead I trimmed the peaks using the analogue El Moo limiter in the McDSP APB system — most limiters are deliberately designed to be transparent, but this one adds a little more character. I then used automation to do the bulk of the vocal levelling.

Choosing vocal effects is always an interesting process, and I’m constantly surprised by how treatments that work perfectly on one voice in one track sound totally out of place in another. I had been wondering whether the vocal reverb in the original mixes was somehow contributing to the sense of things lacking presence; at any rate, I was determined that mine needed to support Robin’s performance without making it sound distant. As is often the case, it seemed easiest to reach this goal by using several effects in parallel. When it comes to vocal reverb, pre‑delay is often the crucial factor, and in this case I used a heavily filtered 77ms delay in SoundToys’ Primal Tap to feed a short plate sound from Arturia’s Plate 140. Augmenting this was a more conventional stereo slapback delay from Wavesfactory’s Echo Cat, and a slightly longer treated delay which I snuck in during the breakdown.

The brass, as previously mentioned, needed some EQ to tame the low mids and, in the case of the trombone, to push the upper mids forward a bit. I also routed both parts to a stereo bus and compressed them as one using IK Multimedia’s White 2A, which helped to reinforce the sense that the section was operating as a single unit — not that they needed a lot of help in that department, as the playing was extremely tight. The brass also got its own plate reverb, with slightly different settings from the vocal ’verb.

Bus Strikes

Like many mixers, I’ll often introduce master bus processing at a fairly early stage of proceedings, but it’s always liable to change. I often find that a master EQ setting works well on everything apart from the drums, and that was definitely the case here. Between them, I had Arturia’s EQ Sitral and Audify’s RZ062A set to add a fair bit of upper midrange and high end, and driving the latter a little introduced some nice colour, but it was all getting too much on the cymbals. What I do in this situation is create two or more sub‑master auxiliaries so that I can separate out the things that benefit from EQ and those that don’t. These are then recombined at the actual master bus for — in this case — saturation from Goodhertz’s Tupe plug‑in, some low‑mid trimming from FabFilter’s Pro‑Q 3, and compression.

On rhythm‑led music like this, where kick and snare are typically the loudest sources, I find it helpful to think of master bus compression as doing one of two things. With a slow attack and fast release, the front end of each drum hit jumps out before the compressor acts, recovering again in time to repeat it for the next hit. This sort of compression thus pulls the drums out of the mix a little. Alternatively, you can dial the attack time right down so that the compressor pushes the drums back into the mix. It’s not always easy to predict which approach will be most effective, but in this case, the second was clearly better. I loved the ‘glue’ that Overloud’s GEM Comp G — an emulation of the notorious SSL G‑series compressor — added with the attack at its fastest setting.

Bus compression from Overloud’s GEM Comp G helped to ‘glue’ everything together, with the fastest possible attack setting serving to push the drums back into the mix slightly.

Down The Line

Choices we make at the recording stage can have consequences that are only felt later on in the production process. There were many good reasons behind Robin’s decision to track at home; it was a space where his musicians felt comfortable, he was set up to video everything, and he has great kit. But working around the space constraints shaped the sound of the resulting recordings in a way that perhaps wouldn’t have happened in a larger live room. The decision to be his own tracking and mixing engineer likewise had obvious benefits, but it also meant I was the first person to hear it who didn’t have a personal investment in the project.

Referencing and mix checking (see box) are invaluable, and there are now many tools that should help identify issues with a mix. But whether it’s a trusted friend, a mastering engineer or the collective wisdom of the SOS Forum, there’s still no substitute for a fresh pair of ears. We all need the reassurance of a safety net, and until AI algorithms and plug‑in presets can provide that, they’ll never replace human beings.

Meanwhile, ‘Ode To NOLA’ has now been released as a single with accompanying video, which you can watch on Robin’s channel at www.youtube.com/repmusic.