This month we deal with thorny phase-cancellation problems, overhaul musical arrangements and get creative with effects, as we rise to the Mix Rescue challenge.
Although Richard Campbell used to play in bands, he's turned more recently to software to realise his musical ideas, using what spare time he has to put together song ideas. He contacted Sound On Sound because he was struggling to flesh these sketches out into finished productions, and sent in a mix of his song, 'Shut Your Eyes', as an example of the results he was achieving.
He had deliberately attempted to use the minimum of equipment, so that he could focus on developing his own style, rather than having to spend too much time learning about lots of kit. His sequencer of choice was Cubase SL 3, which he ran on a Sony Vaio 1.6GHz Pentium M laptop, with a MOTU 828 MkII audio interface. He had used this to create the song from a number of different sound sources.
The main drum loops had been taken from a sample library, but these were supplemented much of the time by a basic, programmed beat, courtesy of Native Instruments Kontakt (v1.5), and an added tambourine part from IK Multimedia's Sampletank 2. The bass sound combined a main, squelchy-filtered patch (which gave the bass its character) and a mellower, deeper layer, for extra 'oomph'. Both sounds came from Rob Papen's Albino 3, while Korg's MS20 (from their Legacy Collection), Cakewalk's Z3ta+ and Novation's V-Station provided some spot effects and textural variety. In addition to the samples and synths, Richard had overdubbed his own Ibanez guitar through a Line 6 Pod to flesh out the choruses in particular, before adding a lead vocal, captured with a Rode NT1 large-diaphragm condenser mic.
Layered Drums: Dealing With Phase Problems
The first thing that needed improvement in Richard's mix was the sound of the drums, which seemed rather soft, especially the kick. A punchier and more up-front sound seemed to me to be more suitable, especially for the song's verses. Listening to the main, sampled loop, it was clear that the equalisation Richard had attempted had been part of the problem, because the raw loop was actually pretty close to the kind of sound I had in mind. A little high-pass filtering at 50Hz just helped tighten it up and left some space for the bass part to manoeuvre at the low end.
These two pairs of audio tracks illustrate the timing problems that Mike encountered early on in the mixing process. Tracks one and three show the waveform of the first kick-drum hit in the sampled loop, but for two different iterations of the sampled loop. Tracks two and four show where the programmed beat from Kontakt had placed its kick-drum alongside each iteration. The way the peaks and troughs of the layered waveforms combined changed with every new hit, and therefore so did the tonality of the composite sound, making it practically impossible to achieve a consistent sound in the mix.
The Kontakt beat also needed very little work in the way of processing to get a suitable sound — just a small amount of de-essing to pull the loop's hi-hat part a little further back in the mix. However, when combining the loops with the programmed beat, I noticed that the kick sound was very inconsistent from hit to hit, even though both parts on their own were fine in this area. Zooming in on the waveforms of the two parts showed that the programmed kick drums weren't always lined up very well with those in the loop, and this was causing phase-cancellation effects, which altered the tonality of the combined kick sound in an unpredictable way.
This is one of the occupational hazards of combining loops and programmed parts, and is the reason why many breaks are severely high-pass-filtered before adding in programmed parts. I didn't really want to high-pass-filter Richard's loops, though, because it would have robbed them of much of their personality. Knowing that the problem would be impossible to deal with using mix processing (I'd have been chasing a moving target), I resolved to spend a bit of time slicing out all the kick drums from the Kontakt drum part, and lining them up manually with the corresponding kick drums in the loop.
With the hits all lined up, the combined kick drums immediately became more consistent and, given that I now had the Kontakt kick drums on a separate mixer channel, I took the opportunity to use some slow-attack compression to emphasise the punch in the sound. The down side of this was that a single cymbal hit at the start of each verse gave the game away by pumping like mad, so I replaced that bar of the pattern with a similar bar that had no cymbal hit, and then added in a nice stereo cymbal sample of my own on a separate track.
Listening critically to the combined sound, I began using some small EQ adjustments to tackle a few little imbalances between the instruments: a notch at 151Hz reduced a low resonance on some of the snare sounds; another notch at 842Hz reined in an over-prominent cowbell part; and a few decibels of gentle low-frequency shelving cut at around 950Hz brought the percussion parts as a whole a step forward. A couple of decibels of peaking cut at 90Hz on the Kontakt drums reduced some unwanted low-frequency build-up and brought the tonality and balance pretty close to what I had been looking for. However, I still wasn't quite happy until I'd dialled in some heavy, saturated compression (from Digital Fishphones Blockfish) to give the sampled loops a bit more grit and attitude overall.
This screenshot shows how audio editing and automation data were used to create more variety in the main sampled drum loop during the first half of the song.The drums were now working well for most of the track, but there was still one last problem to deal with during the song's choruses. Because the sampled drum loop Richard had chosen for these sections had the kick and snare sounds mixed lower in level than they were in the loops elsewhere, this made the verse-chorus transition a bit of a let-down — the opposite of what you'd want. My solution was to slice out the kick and snare hits from the main verse drum-loop and place them on separate tracks alongside the chorus loop, again paying attention to the phase of the slices against the loop hits.
To be honest, this didn't work fantastically well at the outset, but I persevered and ended up processing the extra tracks a fair bit, to carve them into a form where they usefully beefed up the chorus loop. In the first instance this involved some high-pass filtering on both sliced tracks at around 90Hz, as well as a 6dB peaking boost at the same frequency on the kick track. I also squashed both sliced tracks pretty firmly with Blockfish, using a medium attack time, again to emphasise the attack portion of each sound.
The additional tambourine track was fairly straightforward to mix in with the drums, once I'd de-essed it a little to take a bit of the slightly harsh edge off its transients. It seemed a bit lacklustre on its own, though, and the sound was panned off to the left in the stereo track, so I set up a single eighth-note delay, panned to the right, just to give the part a bit more complexity and stereo interest.
Rescued This Month...
"My name is Richard Campbell. I spend a fair portion of my time singing professionally — 'Three Blind Mice' and 'Twinkle Twinkle Little Star' are two of my personal favourites (I work full-time as an infant teacher). I don't have a lot of time for my music, a fate probably common to most bedroom musicians. I have played bass in bands long in the past but now work on my own, purely for my own enjoyment. I have built up a repertoire of half-finished songs and I was keen with the track 'Shut Your Eyes' to finally finish something to my own satisfaction. I was frustrated with the end result and was keen to have someone with more experience than me see if they could do something with my efforts."
+44(0)1926 736875.
Lead Vocals
Now that I was fairly sure that the drums weren't going to stop the mix getting off the ground, I turned my attention to what I considered the most important element of the track— the lead vocal. I often find it makes sense to work like that, because then you tend to fit less important sounds around those that are more important, rather than vice versa. Richard was not at all confident about the quality of his own singing, but I had a hunch that it had mainly suffered from being mixed too far back, especially during the verses. My idea was to have a close-up, crispy verse vocal contrasting with a more spacious sound at the chorus, so I put the different sections on different tracks, to allow separate control over them.
The verse vocal recordings, in particular, were quite muddy-sounding, probably as a result of Richard singing a bit too close to his Rode NT1, and thereby falling foul of the mic's proximity effect. A high-pass filter at 200Hz sorted out most of the problem, but I also cut at around 350Hz with a 1.5-octave-wide peaking filter as well — around 5dB of cut for the verses track, but only 3dB for the choruses track. I decided to go for a highly-compressed sound, partly because it suited the style of the track, and partly because I hoped it would help make the performance feel more assertive. However, I chose the complex opto-compressor mode within Blockfish, so that the compression action itself wouldn't become the star of the show, despite it being driven hard. Unsurprisingly, given the EQ and compressor settings I'd used, sibilance quickly became a problem, so I switched in Digital Fishphones' Spitfish to perform the necessary de-essing.
Bass & Guitars
The bass processing proved to be fairly straightforward. I nailed the lower of the two bass parts with fast, heavy compression to keep it very solid and consistent through the track, and then just high-pass filtered the upper bass part at 100Hz, to avoid the two sounds interfering with each other at sub-bass frequencies. Like most high-resonance synth sounds, the upper bass part had occasional large level peaks that could have had someone's eye out if they'd been left untamed, so I set a high-threshold, high-ratio compressor just to stamp on those, leaving most of the rest of the part comparatively untouched.
The electric guitars were my next stop. The opening guitar part, which also recurs in the second chorus, had been treated to compression, emulated tube-distortion, and some phasing. I quite liked the original sound, but didn't have access to Richard's settings when I began work, so I tried to recreate something akin to it from memory. During this process I tried out various different effects, and more or less stumbled on a nice, grainy sound via a series of Cubase SX2 's Metalizer, Ring Modulator and Overdrive plug-ins. I then compressed it with Blockfish because, although I wanted the timbral modulation effects of my chained plug-ins, I didn't want some of the level-modulation side-effects to stop me placing the sound securely in the mix. I also high-pass-filtered at 400Hz to keep the sound very tightly focused, leaving room for the other parts still to come. When I finally got around to A/B'ing my guitar processing against that in a Cubase file that Richard sent me of his original mix, I ended up preferring the sound I'd got to, so I stuck with it.
Both main vocal parts were processed in a fairly similar manner, with high-pass-filtering, heavy compression, and de-essing. Here you can see the exact settings used.The guitar part that comes in a bit later in the introduction also plays in both choruses, and this needed very little work at all. Again, I high-pass-filtered it, as it had masses of low end, which just obscured the kick and bass to no good purpose. This time I used a cutoff frequency of around 225Hz, as well as a 6dB high-shelving cut at 13kHz, in order to leave it sounding warmer in contrast with the other guitar.I opposition-panned these two guitar parts in the second chorus, but found that this left the stereo image in the first chorus sounding a bit imbalanced with only the one guitar. I could have copied the other guitar part back from the second chorus, but I was worried that it would leave little room for the second chorus to build on the first. In the end I used a similar delay treatment to the one I'd used on the tambourine track, to generate an eighth-note delay of the first guitar part on the other side of the stereo picture. The send to this effect was set so that it just helped balance the stereo picture of the first chorus, without subsequently imbalancing the second chorus.
For further contrast between the two guitar parts, and also an increased sense of front-back perspective, I dialled in a little reverb from the SIR convolution plug-in. Given the nature of the track I didn't want anything too smooth, so instead I started with a fairly bright impulse response of just over a second in length, and then adjusted the pre-delay and impulse envelope controls to make it into something more like a diffuse slapback — a slightly indie 'in a garage' sound. I high-pass-filtered the reverb return, as usual, to avoid low-end clutter, and then set a fairly subliminal level.
The heavily-processed guitar parts that enter during the second half of the second verse were also processed along the same lines as in Richard's original mix, but without his exact settings. I felt that something a bit wider than the other very clearly-focused sounds would create a nice contrast at that point in the track, so I panned the tracks a little left and right before sending them to a communal stereo bus for further processing.
I could hear that Cubase' s Step Filter plug-in had been used in Richard's mix, but I was pretty certain I also wanted this part to be pretty thin and 'watery' to fit around what I already had going, so I went through the Flanger and Symphonic plug-ins first. While trying out a few different things, I found that driving the Magneto tape-saturation plug-in hard added something worthwhile, so I chucked that in as well. Finally, I added an instance of the Chopper plug-in, using a very slow tempo-sync'd sawtooth-wave modulation, to give a slight level swell through each bar. This is the kind of thing I used to do in the past by keying slow-attack duckers, and it's a great technique for surreptitiously emphasising any track's bar-to-bar momentum.
Hear The Differences For Yourself!
The following audio files demonstrate some of the mix processing I used to create my remix — you can listen to or download these at www.soundonsound.com/sos/sep07/articles/mixrescueaudio.htm.
layereddrumsraw-0907.mp3
Synth Processing
The synths still remained to be mixed in, but this didn't turn out to be too tricky. A lot of synth sounds, especially ones that are heavily modulated or use high-resonance filtering, can be very difficult to balance with other parts, because their frequency responses vary so wildly from moment to moment. However, a little multi-band compression quickly makes sounds like these easier to manage in this respect, and such processing doesn't usually change the essential timbral quality as much as you might think. I put this tactic into practice for the squelchy synth part that enters in the second chorus, as well as for the stereo arpeggiated synth before the second chorus, whereupon it became pretty straightforward to find appropriate levels for them both. I also tweaked the output levels of the frequency bands to tone down the fizziness from the arpeggiated synth — it was a little much for my taste.
In SIR, the pre-delay and impulse envelope controls were used to create a diffuse slapback sound for one of the two main guitar parts during the intro and second chorus. Some high-pass filtering was also used to keep the low end of the mix nice and clear.
The other main synth part was the filter 'burbling' under the closing sections of the song. This had a lot of sub-bass and, while it was important for the sound to have a full frequency range, I was worried about subsonic elements eating into the mix headroom for no real benefit, so I high-pass-filtered the track at 30Hz. Other than this I didn't see the need for any other EQ at all. I liked the inherent dynamics of the part, and the way these fitted within the arrangement, although I did use the Soft Clip and Limiter sections of Cubase 's VST Dynamics plug-in to keep a lid on a few over-enthusiastic level peaks.
Richard had put a fairly nondescript pad-synth sound behind the second chorus, presumably to give it a little extra warmth compared with the first chorus. With the guitar sounds and effects now fulfilling that role, it had become a little redundant, so I decided to see if I could coax it into a different role, as a high counter-melody. This involved twice high-pass-filtering the sound at around 2kHz, leaving only those frequencies that would float over the rest of the tracks without clouding the mix's mid-range. On its own, this synth now sounded almost offensively thin, but against the whole mix the miracles of frequency masking left it sounding pretty appropriate. I still used a little phasing to mellow the sound a bit, along with some compression to counteract some unwanted level modulation in the synth patch itself.
By this point I had most of the mix fitting together sonically (plus or minus a few effects and automation rides), but I found myself nevertheless becoming increasingly dissatisfied with the overall result. In the end I had to admit that I just wasn't convinced that the arrangement was providing enough variety to sustain the interest of the listener, however peachy I might be able to make the sonics. After much dithering on my part, I decided to bite the bullet, step a bit outside the normal Mix Rescue brief, and have a crack at remodelling the track's arrangement. Although I knew there was the risk that Richard might hate me chopping his material around, I simply couldn't see a way to make the track work through mix polishing alone.
Beyond Mix Engineering
Richard mentioned in his initial letter to SOS that he felt that the quality of his initial recordings was letting down his mix, and this is something I hear time and again from home studio owners. Although such feelings are often justified, it is also just as common for a track's arrangement to compromise its final presentation, but this is rarely touched on in most discussions of mixing. One of the reasons why many commercial albums are mixed by independent engineers is that a fresh view of the material at the mixing stage can really help to focus the arrangement, not least through ruthless use of the console's mute buttons! So if your mix isn't really reaching escape velocity, dust off your sequencer's scissors tool and try carving things up a bit.
I also hope that this particular Mix Rescue example will encourage more of you to work with your effects in a more dynamic way — not just setting and forgetting. Special effects really come to life with a bit of automation data thrown at them, but even apparently background treatments like delay and reverb often do their job better if send and return levels are tailored to suit different sections of a song.
Arrangement Enhancements
Initially I just looked for areas of different parts that could simply be muted, allowing otherwise buried arrangement details to shine through. On a small scale, this involved things like muting all parts other than the main drum loop during the last fill before the first verse, or dropping out everything but the vocals, tambourine and bass in the middle of the second verse. When there are a lot of complex sounds layered together in a song, I find it really helps me as a listener to notice them all if each one gets at least a moment in the spotlight. Muting small sections of audio was also useful for creating more arrangement variety in some sections, such as during the bars before the first chorus.
On a slightly larger scale, I also muted some track sections to help with the overall mix dynamics. For instance, I killed both the deeper bass sound and the sliced Kontakt snare at the start of verse two, giving me room to build up the texture through that verse as I reintroduced them. To my mind, another beneficial side-effect of this is that it clearly refocuses attention back onto the lead vocal after the fuller texture of the preceding chorus.
In other places I began to get a bit more creative with the editing techniques. For example, reversing bits of audio turned out to work quite well for the guitar chord just before chorus one and for the last bar of the arpeggiated synth part before chorus two. Things got a bit more complicated for the guitar part after the second chorus, which was one of the areas I initially felt least happy with — it seemed as though the track was just treading water, and I almost cut the whole section out of the song for that reason. I stayed execution for a little while, though, to see if I could come up with something more engaging by chopping the whole part up into 16th-note chunks and then creating a rhythm by muting some of them. Once I had a rhythm I liked, I processed some of the slices off-line with Cubase 's Tranceformer and Grungelizer plug-ins, and also created pitch sweeps for some slices with Cubase 's envelope-controlled off-line pitch-shifting. As a final touch, I panned the part left and then sent it to a trippy 3/16th-note delay effect on the other side of the stereo field, by which point I felt that the section now merited its place in the song.
Remix Reactions
Richard Campbell: "I was very eager for my track to be considered for Mix Rescue, because I felt my original efforts at mixing could hardly be called that! I applied effects and altered tracks, but before long I felt lost in a sprawling mix that was supposed to be simple and stripped back.
"To say that I was impressed when I heard what Mike had done to the track does not adequately cover it. There were a number of things that slapped me in the face after the initial excitement of hearing how the new improved version sounded. I now felt good about hearing my own voice. It seems much more smooth and able to stand up in the places it is more exposed in the song, and despite being breathy and very upfront, it still blends with the track. This was the major area that had been frustrating me. The bass also seems much warmer, more solid, and overall just much more pleasant to listen to."The biggest difference to me, however, was the change that Mike made to the arrangement. I am very pleased with the direction he took, and he has made the song flow much more smoothly. It now has the punch I was looking for in vain, as well as the fairy-dust sprinkled over individual sounds. I did not even realise that the synth sound in the second chorus was my original sound! The effects now seem to tie the whole song together and make it really sound finished, instead of just like a demo.
"I can't say enough how good I think the track sounds now! Mike has made my quick, simple song into one I can be proud of. It sounds 'produced', but in a way that was sympathetic to my original idea — I had a sound in my head which I wanted to get out, and Mike managed to get there for me. Now I have something to aim for when I try my own mixes."
Creative Effects
My other main strategy for spicing up the sound was using more off-the-wall effects. I like to use tempo-sync'd delay plug-ins for blending together sounds in my mixes, so that I don't have to use too much reverb (which can leave a mix a bit 'washy' for my preference), and this mix was no different in this regard. However, I deliberately processed a number of the delay returns in this situation, to make them slightly less of a background feature.
So, for example, a fairly unremarkable eighth-note delay patch I had decided to use to give the chorus vocals a more spacious character was further processed, with a slow, overdriven rotary-speaker effect, while a delay effect on the second guitar part in the introduction was converted into a stream of 16th-note pulses by the Chopper plug-in, before being passed to Cubase 's Rotary and Symphonic effects. This latter effect sounds almost like a synth part, and I liked it enough that I also fed it from one of the chorus guitars and faded it in for a number of sections in the second half of the song. I later took advantage of having several different delays available, and threw pretty much all of them at the burbling synth part during the outro, to keep the stereo interest going and give a bit more of a rhythmic feel.
This screenshot shows all the different send effects used in the final mix. Five of these were delay effects, while the other two were reverb from the convolution plug-in SIR and a Symphonic effect that was used sparingly on some of the lead vocals.
Richard hadn't done a bad job of recording his vocals, but they were a little muddy sounding — probably due to the proximity effect caused by singing too close to the mic — and needed EQ to compensate for this.Automation played a large part in the implementation of the send effects in this mix. In some places I used gentle automation ramps to implement slowly evolving changes (for example, where the guitar's pulsed delay effect slowly builds into the final drum-fill before verse one). Unnatural muting of the effect returns helped make the dryness of the verses particularly stark, while the sudden addition of a variety of delay effects after the second chorus gave extra width and complexity to the guitar chord and synth arpeggiations. Periodically increasing the lead vocal's delay level and feedback really helped to sustain the end of each line of the chorus nicely, and it also contributed to the slow, evolving fade-out of the outro.
Insert plug-ins played their part as well, supplying a number of momentary 'spot effects'. A couple of examples of this occur during the lead-in to the first chorus. Here I made a copy of the little "your anger" phrase, and used automation to switch in both a Step Filter insert effect and a delay send to the chorus vocal's delay effect. A heavily-overdriven Tonic high-pass filter plug-in was switched in over the main drum loop, performing a wide filter sweep.
I re-used that same Tonic plug-in elsewhere on the drum-loop track as well: there's a high-resonance low-pass filter sweep midway through verse one (complementing a high-pass filter sweep I inserted on the simultaneous synth fill), another one before chorus two, and then a slow band-pass sweep during the outro. A final instance of Tonic helped out on the guitar chord during the instrumental section between the second verse and chorus — I'd already looped this chord using Cubase 's audio editing tools, ending it with a reversed section to crescendo into the chorus. During the looped section of the chord, I switched in Tonic as an insert effect and then automated its mix control so that a sample-and-hold-modulated filter faded in and out. Alongside this I also layered a couple of extra sound-effect samples (a string scrape and some tuned telecom noises), as the song still seemed to me to need a couple of fresh sounds at this point.
There were many more of these kinds of 'nips and tucks' than I've covered in the preceding paragraphs, but hopefully I've been able to give an idea of the main tricks I used. What is less easy to communicate in print is how much refining and tweaking is involved in pulling so many little effects together into a coherent whole. It's very much a juggling act, where you're constantly moving between different plug-ins, automation lanes and audio-editing tools, and I had to revisit many of my edits and effects automation curves a number of times before they were working satisfactorily within the context of the mix as a whole.
With so many changes to the musical elements of the track, I was getting a bit worried that I might be seriously misinterpreting the direction Richard wanted me to head in, so as soon as I had the first rearranged mix hanging together roughly, I decided that I should seek his feedback, emailing him an MP3. Fortunately he seemed pretty happy with the direction I had been going in, which gave me extra confidence to push on with some of my more left-field processing choices, such as the chopped guitar parts after chorus two. After sorting those out, it was mainly just a question of finessing the lead vocal levels (in order to keep audibility throughout all the arrangement and effect changes), before the job was done.