It pays to keep your ears open for phase problems when mixing. Tackling them means you shouldn't have to resort to radical EQ so often.
I recently judged a big mixing competition at www.mixoff.org, during the process of which a small‑studio multitrack recording from the Australian three‑piece rock band Young Griffo was presented in more than 100 different mix versions! In the process, I wrote detailed critiques of most of these mixes to help the entrants improve their work, and once the competition was complete, I posted all of them (along with all the mixes and the original multitrack download) onto a single archive page at www.cambridge-mt.com/YoungGriffoCompetition.htm, to create a more convenient educational resource.
What I didn't do, though, was actually mix the multitrack myself to demonstrate how I would have tackled those same recordings in practice. Although I was actually tempted to have a go, simply because I loved the song (even after two weeks solid listening to it!), fate intervened when the band mentioned their disappointment with a commercial mix they'd received. Although they had managed to put together the majority of an EP in collaboration with an excellent local semi‑pro mixer, he was unavailable to work on the final song, 'Facade', for them, so they spent the remainder of their slimline budget to hire a mix engineer to fill the breach. Unfortunately, things turned out very differently than they'd hoped. "We were shocked by the poor quality of mixes we received from the pro mixer,” they told me. "Clearly, minimal effort had been put into the mix. A completely lifeless drum sound, a round and muddy bass tone, bland and flanged‑out guitars (we didn't ask for flange, but man, we got it!), disembodied vocals, and an all‑round flat sound. In fact, we found ourselves considering the quick rough mix from the tracking sessions to be a slightly better standard! So in the end we were left with a demo‑standard final track.”
In the circumstances, I therefore suggested that I try to provide a more serviceable remix for them, and in the process provide tips and tricks for dealing with all the same common mix problems that contestants had struggled with during the mix competition, given that both multitracks had been recorded in a very similar fashion.
The first big thing that tripped entrants up in one way or another was phase‑cancellation, because everything except the vocals had been multi‑miked during the recording process. If you record the same instrument through several mics simultaneously, time delays between their recorded signals mean that you can seldom simply mix together the sonic characteristics of each as you might hope. This is because the signals actually interfere (or 'phase‑cancel') to some extent, introducing potentially dramatic ripples into the combined sound's frequency response. For more details on why this happens, have a look at the 'Phase Demystified' feature back in SOS April 2008: /sos/apr08/articles/phasedemystified.htm.
The guitars probably provided the simplest practical scenario, so let's look at how I dealt with those first. Each of the two guitar performances (the main rhythm part and an overdubbed solo) had been played through Vox and Orange amps simultaneously, the former captured with Beyerdynamic M160 and Shure SM57 mics, the latter with a Sennheiser MD421 and a Cascade Fat Head ribbon. Having so many mics to play with is great from a tonal perspective, because it allows you masses of scope to mould your guitar tone without having to use EQ. In fact, you can even turn the often undesirable frequency artifacts of phase‑cancellation to your advantage to help find the timbre you're looking for.
However, if you do decide to allow significant phase‑cancellation between the microphones, and then pan them right across the stereo field, the frequency effects of phase‑cancellation become much less apparent because the conflicting signals are coming from different speaker/headphone drivers. That might seem fine and dandy when you're listening in stereo, but if anyone plays your mix in mono (as they will in a lot of real‑world playback environments) the phase‑cancellations have nowhere to hide and your guitar tone may completely collapse! In a nutshell, then, either you've got to minimise phase‑cancellation between the mic signals, or you've got to pan all the mics roughly to the same position.
In the case of a high‑power three‑piece such as Young Griffo, there's a strong temptation to pan the mics to opposite corners of the soundstage, in order to get the single guitar part sounding as enormous as possible. The moment you do this, though, the desire for mono compatibility restricts your freedom to use creative phase‑cancellation between the mics to mould the instrument's timbre. So do you settle for less width, or less tonal control?
As with most mixing decisions, it helps to have some idea of the demands of the style before coming to any firm conclusion, and fortunately the band had provided me with a handful of commercial productions for reference purposes, much as they had done for the mix contest. The main one was 'All The World Is Mad' by the band Thrice, although Cog's 'Are You Interested' had also been mentioned specifically in relation to guitar sounds, and these clearly pointed the way ahead, because they both used wide‑panned, double‑tracked rhythm parts to create their wide stereo image. My solution to the 'Facade' quandary, therefore, was to edit together a rhythm‑part double‑track by copying and pasting the guitar parts between the two choruses (something that's a cinch to do with any repetitive part like this), allowing me to keep the mic signals for each guitar part panned close together on opposite sides of the stereo picture. That meant that I could freely adjust the guitar tone without unduly compromising the mono compatibility of the track as a whole.
It was handy to have that extra tonal flexibility, too, because (as is common on low‑budget productions) the guitar part in question had been overdriven a touch too hard, so it took some careful work to bring about a mix sound that was solid and present, but without too much 4‑6kHz abrasiveness or loss of note definition. I started by trying to find the mic that had the most appealing mid-range tone, which turned out to be the M160. There was some roughness in the high frequencies to attend to with 6dB peaking cuts at 5.2kHz and 11.3kHz from Cockos ReaEQ, but once that had been tamed, the only other EQ I applied was to give the sound a bit more 'centre': 2.5dB of 670Hz peaking boost and 2dB of 8kHz shelving cut from Sonimus SonEQ, a great new freeware Pultec/API EQ modeller I've recently discovered.
The SM57 seemed the next most promising track, although I quickly filtered of some excess woof and fizz from it with ReaEQ's high‑pass and low‑pass filters, to bolster the M160's mid-range. Experimentation with different levels, polarity settings and time delays for the two tracks rewarded me with some useful added 'body', but I decided to push on for bonus points too, inserting an instance of Audiocation's excellent Phase plug‑in to finesse the result with a significant reduction in the unwanted HF edginess.
I was becoming pretty satisfied with the guitar sound, but still tried out the other mics to see if there was anything to be gained. The Orange amp's MD421 didn't provide anything different, no matter how I massaged the time‑alignment, polarity or phase relationships, but the Fat Head seemed to offer a fraction more warmth and presence when nudged into the mix with some of its low-mid range carved away, and another instance of Audiocation's Phase applied for good measure.
The guitar sound, then, was defined as much by the balance, polarity, timing and phase settings as it was by my EQ, and this is very much par for the course: there's no point in over-using EQ while all but ignoring the equally important 'suck it and see' timing/phase/polarity tweaks. The same goes when processing a bass guitar that has been captured with a miked amp and a DI simultaneously, as Young Griffo had done. Although my processing in this instance involved a couple of enormous EQ peaks (the first time I can remember maxing out a boost control in Stillwell Audio's gutsy Vibe EQ plug‑in!), to approach the Death Cab For Cutie bass sound that the band had their eye on, time‑aligning the mic and DI signals also proved to be vital in achieving a reliable low mid-range tone.
I've discussed issues of phase as related to drum mixing on numerous occasions, so I won't go over that ground again here, despite its importance for this final mix. Instead, I'd like to focus on one aspect of the drum that foxed a large number of the competition entrants: achieving an effective snare sound. The nub of the issue is that a snare in any heavy-rock texture needs to have serious sustain and density if it's to avoid being turned into an uninspiring 'blip' by the frequency‑masking effects of the wall of guitars. However, the difficulty with achieving this, as far as most small‑studio drum recordings are concerned, at least, is that the only mic that's picking up much of the snare is positioned a couple of inches away from it, and thus usually gives you very little to work with — it's all stick attack and over‑emphasised pitched resonances.
If your snare part is simple, the quickest solution to this is simply to draft in a sample, but I was reluctant to do this with a more intricate performance such as the one in 'Facade', because I find it can easily start sounding very synthetic. So I concentrated instead on making the best of the close mic, using a combination of compression, distortion and reverb. The compression was fairly brutal, using an instance of Audiocation's AC1 plug‑in at an 8:1 ratio to knock 12dB off each hit's attack, thereby dredging up a lot more of the instrument's decay tail. Attack and release settings are pretty critical with such extreme gain reduction, and I found myself revisiting these on a number of occasions to fine‑tune the remaining attack character and the envelope of the enhanced sustain — 33 microseconds and 8ms were the final settings, respectively.
Complicating this channel's processing, though, was the fact that the level of hi‑hat spill on the snare mic was already quite high, and was therefore being seriously over‑emphasised by my weapons‑grade compression setting. Gating the snare tight enough to remedy this for the whole song made the snare come across as rather robotic and sampled, though. My solution was to automate the gate's release control so that snare hits were only closed down fast if absolutely necessary, thus preventing closely following hat strikes from breaking through.
Additional harmonic density came from Digital Fishphones' rather long‑in‑the‑tooth freeware THD plug‑in, operating in soft‑clipping mode, and the under‑snare close mic also had a role to play here, thickening the 1‑4kHz region in particular with its burst of snare‑wire buzz. Finally, a couple of different reverb treatments further extended the instrument's length: the first was an impulse response of a resonating piano running in Christian Knufinke's SIR2 convolution engine, specifically chosen to add some clangy, garage‑rock resonances to the upper spectrum; and the second was a longer reverb from Antress Modern's no‑frills freeware Flashverb plug‑in, which also added some stereo width. In neither case was the treatment very realistic, in the sense of sounding like a natural room, but as tonal and sustain enhancements for the snare, they nonetheless did the job handsomely.
There was a lot of discussion during the mix contest about how much competitors should be allowed (or expected) to tamper with the supplied multitracks. On one side of the fence were entrants who felt that the raw recording should be mixed as it stood, as a true reflection of the band's intentions and sound, while other contestants felt that any amount of editing and overdubbing was fair game to win the band's approval. As will be clear to anyone who's followed my Mix Rescue columns in the past, I'm more in the latter camp, as it's my experience that many self‑produced recording musicians don't realise many of the things that go into making the commercial records that they aspire to. There's a balance to be struck, though, because it's pretty obvious that no‑nonsense live acts such as Young Griffo are unlikely to warm to a mix that dramatically alters their musical material or makes them sound like some kind of studio confection. So although I used a number of edits and additional sounds to bolster my remix, I went to some lengths to avoid these coming across too much like 'artificial additives'.
For instance, in order to fill out the vocal sound, I created another 'fake' double‑track by editing between the two choruses. However, ensuring that the double‑track was very tightly in sync with the lead part — and at a fairly low level, too — kept it from undermining the sense of a single voice singing most of the time. I pulled a similar stunt with the chorus section's bass line, doubling it with some double‑tracked guitar power-chords from Nine Volt Audio's Big Bad Guitars sample library. Again, though, I didn't want these additions to sound like extra parts, so edited them very tightly to follow the bass line, and then tucked them out of the way behind the main rhythm guitars, using high‑frequency EQ cuts from Cockos ReaEQ and stereo-width reduction from Voxengo MSED. As a result, it doesn't sound like an independent overdub, but just makes the band's bass and rhythm‑guitar parts sound fuller and more powerful.
A similar principle underpinned my use of additional cymbal samples over the drums. The problem I encountered was that the stereo overheads recording was rather phasey‑sounding — the result, I suspect, of using a Blumlein stereo mic array too close to the kit. Although I was able to improve the mono compatibility of the raw overheads a little, using Audiocation's Phase plug‑in in the left channel, I still wasn't happy with the way they translated to mono, so I narrowed them in stereo to reduce the mismatch, a process that brought with it some undesirable tonal thinning. The purpose of the added samples (from Spectrasonics's Backbeat library) was simply to reinstate some of the cymbals' lost width and tone — and, by implication, to improve the apparent recording fidelity of the entire kit. This was especially important given that the band's drummer had cited the cymbal‑rich kit sound of Biffy Clyro's 'That Golden Rule' as a reference point.
One of the great challenges of a song like 'Facade' is maximising the contrast between verses and choruses. While you want there to be some aspects of the sonics that unify the whole production, you simply can't expect that mixer settings that work for one section will work for them all, so multing and automation are indispensable if you're looking to compete with commercial releases in this kind of style. What you mult and what you automate is primarily a question of what makes most practical sense. In this instance, I multed the guitars and vocals to allow completely different EQ and effects settings, whereas I used automation for the drums and bass, to implement a range of different adjustments on different channels. The kick and snare received the lion's share of the automation work, with additional high‑frequency energy being added to aid cut‑through against the powerful chorus guitars, and a series of rides to adjust the contributions from the under-snare and Flashverb channels to the overall snare timbre. The 2‑6kHz region of the added cymbal samples was also carved away during the choruses, to avoid a build‑up of harshness there. Pushing up the bass amp track by a couple of decibels gave another little lift there, too.
Beyond the longer‑term section tweaks, I also put in a few hours' work riding the lead vocal, snare and guitar levels throughout the track, to try to keep the balance stable, to maintain good intelligibility for the lyrics, and to draw greater attention to some nice backing details. This was another area that was generally undervalued by the competition contestants, and while it's not exactly glamorous work, it makes a big difference to how the music is received on an emotional level.
There can be a lot of work involved in making a rock production sound powerful, especially if the arrangement is ostensibly quite sparse. Keeping on top of phase and masking issues will take you a good part of the way towards a finished result, as will careful corrective editing, but adding supplementary audio parts can also bring considerable benefits if you are able to do so in a sensitive and sympathetic way. Whatever your views on that, though, you'll struggle to reach a truly commercial sound these days without taking full advantage of your DAW's automation system, so don't forget to factor that into your workflow sufficiently.
Bus compression features heavily in most rock mixing, but surprisingly few of the competition entrants pushed this aspect of the sound very far. In the 'Facade' remix, I used a combination of two stages of bus compression, both courtesy of Stillwell Audio's flexible Bombardier plug‑in. Firstly I knocked 3‑6dB off the overall drums bus at a 5:1 ratio, setting the a 20ms attack time to avoid losing too much attack or low end, and then dialling in around 100ms of release time by ear, to achieve a nice pumping effect on the cymbals. I then applied a further layer of slow‑release 2:1 compression on the master output bus, to provide more in the way of subtle mix cohesion.
In addition to that, I inserted an instance of the new Universal Audio Studer A800 plug‑in across the master bus. This was something I actually did quite early on in the mixing process to counteract a slightly gritty high‑frequency character which seemed to pervade the recordings as a whole, especially the drums and the guitars. Some engineers would recommend leaving this kind of processing to the mastering stage, but my feeling is that tape emulation is a complex dynamic process and, like other master‑buss dynamics processing, will almost inevitably affect your mix‑processing and balance decisions, so I decided to mix through it so that I could make more informed decisions about the final sound.
Young Griffo are an indie rock band from Brisbane, comprising Mat Gilroy (vocals, bass), Todd Orchard (Drums), and SOS reader Paul Craig (Guitar). Founded in 2009, they've gigged regularly since, developing their own powerful sound influenced by bands such as Manchester Orchestra, Death Cab For Cutie, and Biffy Clyro, They recently reached the number one spot on Triple J radio's unsigned indie chart with their song 'Pennies', and are planning to release their first EP later this year.
Paul Craig: "This mix is outstanding, we all listened to it and were blown away by it! The original mix had left us questioning whether we should just scrap the song and start again, but Mike has now returned it to us with natural warm vocals, 'live' sounding guitars, a grinding bass tone, and punchy‑sounding drums, so we're now really excited about this track being on the EP. Everything just works really well together, without individual instruments standing out — no hero elements, just three instruments and a vocal pushing the song along together. Mike's also nailed the contrasts required between the verses and choruses, so that the verses stumble along almost carelessly with a perfectly sitting vocal/instrumental balance, before that chorus kicks in and blows your head off — in a good way! The minimal mixing effects are exactly what we wanted too, leaving just a great, honest mix.
"The heart of what we do as a band is our live show, and the best thing we can say is that this mix now captures that energy. It's the best way we know to represent ourselves as a band and we are thrilled to hear such a dynamic and passionate‑sounding version of 'Facade'. We can't thank Mike enough for his efforts — he's done an amazing job and gone far beyond where we ever expected this track to be.”
We've placed a number of 'before and after' audio files, both of the full mix and of the various tracks it comprises, on the SOS web site, so you can download them and listen to the differences in your own DAW. Go to /sos/oct11/articles/mixrescuemedia.htm for more.