Beefing up bass parts played on guitar and importing GarageBand songs into Logic are among the challenges for Mix Rescue, as we add some spice to Adam Bevan's mix.
Adam Bevan is a university student (no, not music technology!) who records his own songs at home, using a very modest system based around GarageBand and running on a Mac laptop. His song 'Deserts' comprises mainly 'real' instruments and voices (all played and sung by Adam), but the rhythm comes courtesy of Apple Loops drum samples.
The version of the song Adam gave me was imaginatively structured, and included a processed but clean-sounding electric guitar solo over a rhythm groove during the intro, some fairly gritty rhythm guitars during the main sections, and a very effective chorus — where everything drops out other than the drums and an acoustic guitar, allowing the three‑part vocals to come through strongly. Adam hadn't bought a bass guitar at the time he made this recording, so he used his electric guitar and then tried to process it to make it sound more like bass, using GarageBand's Guitar Amp plug‑in and some EQ. Although tonally successful, the resulting sound inevitably lacked the low‑frequency punch of a real bass.
The song also had the potential for strong dynamics, with extra rhythm layers and additional guitar parts coming in towards the end, but the mix itself was starting to get muddy. Adam had used distortion plug‑ins on some of his vocal parts to create an obviously effected sound, but this also detracted from their clarity, and covered up the fact that he really has a very good voice. He had spread his vocal parts over several tracks so that he could apply different effects to the different parts, and he'd also tried the same trick with the guitars, to introduce some tonal variety throughout the song. Part of his difficulty in creating a strong but uncluttered mix was that many of the sounds occupied the same part of the audio spectrum, so that the fuzzy guitars, treated 'fake' bass and fuzzy vocals tended to blur into one.
Before working on the song, I imported it into Logic Pro 8. It's not something I've needed to do before, but the process turned out to be pretty straightforward. Still, I did find that there are a couple of things that you need to watch out for...
Adam first brought me the mixes on a Flash drive, but he hadn't checked the 'Save Assets' option when saving the GarageBand project, which meant that if I didn't have the same Apple Loops as he did, they wouldn't play back. Furthermore, for some reason neither of us could fathom, every mix that we imported had all the channels routed to bus 11, which was then routed to the stereo mix. They played without problem, of course, but it was only a matter of moments to redirect the channels to the main stereo mix output.
Prior to starting the 'real' work, it was necessary to trim the various audio regions, especially the distorted guitar parts, to ensure that there was no hum or buzz preceding them, and I also used Logic's fade tool to fade out the ends, ensuring a smooth transition to silence.
Adam had used aux sends correctly, and had set one up to feed an instance of Platinumverb, and another to feed a delay effect. Other than swapping Platinumverb for a short, bright EMT plate (courtesy of Univeral Audio's Plate 140 UAD plug‑in), I left these as they were, adjusting only the levels.
Curiously, any plug‑ins used within GarageBand show up as a plain panel of controls in Logic, rather than the more friendly Edit view, so all these had to be switched over manually. Another curiosity was that when switching on the low‑pass filter in the channel EQ, the frequency always defaulted to 20Hz — rather than the more logical 20kHz — so that had to be reset manually as well.
These were all very minor irritations, but rather more serious was that the audio files imported from GarageBand didn't seem to allow the region-enlarging function to be used in Logic's Arrange Page. To get around this I found that I had to do an unnecessary edit, such as adding a bit of silence to the end of the audio taken from elsewhere, then using the glue tool to create a new file. The new Logic-format audio file that was created as a result could then be extended as normal, where required.
Before fine‑tuning the sounds, I looked at the song structure and identified a repetitive instrumental section that felt as though it went on a bit too long without developing. Some of the breaks also sounded a bit unfinished, partly because Adam had simply used level automation to stop the drum loops from playing when he needed a break, but there were no ending or starting drum fills to create the illusion of a real performance. Often, all that was needed was an extra pasted beat with a cymbal crash layered on top, but in other areas more sophisticated solutions had to be found.
I shortened the rather repetitive middle instrumental section by four bars, and also trimmed the end of the song, where I felt the solo guitar outro went on a bit too long before fading out. Adam had sung a long, sustained note over the first half of the instrumental section, and I stretched the tail end of it quite considerably, to extend over the entire length of the section, gradually fading it out and also putting in some automated panning.
I also took the liberty of reshaping the two‑bar break after the solo, using chopped-up pieces of Adam's original guitar parts, including two heavily stretched guitar sounds that now sound more like a car skid and crash! There's a limit to how far Logic allows you to stretch a sound in one go, so I repeated the process two or three times to move the sound as far away from reality as possible. One short chord became a drawn‑out screech — but, crucially, it still retained its pitch. Those stretched guitar sounds were dropped in again towards the end of the song, just as the solo'ed rotary guitar outro takes over, giving more of a sense of finality.
During the first half of the break I also alternated two very short sections of guitar (less than one beat long and pasted four to the bar), copied from Adam's previous parts, then panned them left and right, fading their levels over the course of two bars. An automated low‑pass filter was also used on these same chordal stabs, to progressively dull down the sound as it repeated. To add interest and impact, I brought the song back in after the break using a trio of very short guitar clips (again, copied from earlier guitar tracks) but edited to sound like stabs or stutters, followed by a simple snare‑drum fill to get back into the rhythm loop that was about to start up. Adam's solo bass part during the break also didn't seem to fit as well as it might, so I opted to take just half a bar of his bass part and loop it four times.
Adam also hadn't used any panning, because he was unsure what approach to use, so I decided to try to get the mix sounding good in mono first, before using panning to separate the layered guitar parts and layered vocals. Now it was time to work on the sounds themselves.
I noticed that although Adam had inserted compressor presets on various guitar tracks, he hadn't adjusted the threshold, which meant that on some of them there was no gain‑reduction taking place at all. Heavily distorted guitars tend not to need compression anyway, so I simply removed any unnecessary compressors.
Adam wanted an aggressive, Slipknot‑style guitar tone, but had DI'd the guitar and was not actually playing in a very aggressive way, so getting the necessary angst in there proved to be a bit of a challenge. For his main guitar part, I took out the Guitar Amp Pro plug‑in and instead tried Line 6's Pod Farm, running a Bomber XTC model with some EQ lift at 1.2 kHz and 100Hz. I also fell back on my old trick of using a tremolo plug‑in, adjusted to work as a chopper (sixteenths), to add a sense of rhythm to one of the guitar parts, but then adjusted the modulation depth so that the effect was a little more subtle than usual. As Adam wanted the guitar to be reasonably loud, I also configured a noise-gate plug‑in to work as a ducker triggered from the main vocal, so that the level would drop by 2dB when Adam was singing. This wasn't obtrusive in the finished mix but it did give the vocals a bit more room to breathe.
For another heavy guitar part I tried a different Pod Farm amp, based on a well‑known British stack, in conjunction with a very mild dose of Logic's Phase Distortion plug‑in patched before it, which helped to get a 'nastier', more aggressive sound, with some sense of 'death metal' intermodulation and grind.
In general, Adam had been tempted to add quite high levels of distortion, but this can actually detract from the impact of the sound if you go too far, and all those extra harmonics spread right across the audio spectrum made it hard to create any separation between the guitars. The distorted guitar tracks responded well to some 'bracketing' EQ using 18 or 24dB/octave high‑ and low‑cut filters to remove the fizzy high‑end and boomy low‑end. In Logic's Channel EQ, the low‑pass filter also has a resonance control, which was useful in peaking up the upper mid‑range of the guitar before the cut came in as, in conjunction with some gentle EQ boost in the 3kHz region, it helped enhance the 'bite' part of the guitar spectrum.
The acoustic guitar part was treated using the Waves Maserati Acoustic Guitar plug‑in, followed by some extra compression to thin out the sound and create some ambience, while the chorus electric guitar that brings the song in, and again reappears at the end, was changed to a slow rotary speaker effect.
One thing I learned from Adam's mix was that he likes variety, and he doesn't like the voices or guitars to sound the same all the way through the song. He'd tried various tricks, such as changing the amount of reverb or delay, but he'd also experimented again with vocal distortion.
Simple distortion plug‑ins tend to make vocal parts sound fuzzy and rather messy, so I tried SPL's Twin Tube plug‑in on the main vocal, and added just a bit of thickness to it that way. This gave a tube‑like warmth and a bit of growly distortion — but without the fizz. Prior to the Twin Tube I used a Universal Audio LA2A compressor to level the vocal and give it a bit of punch, and a Waves Maserati Vocal plug‑in (which combines compression, EQ and delay effects) gave the part more of an up‑front sound. Because this plug‑in can add both delay and reverb, no additional ambience effects were required.
As Adam wanted a double‑tracked kind of sound, I first dusted off my old ADT trick, using Logic's Platinumverb, and following it with an instance of the Antares AVOX doubler to further thicken the ADT effect. Platinumverb has a balance control for the reverb tail and the early reflections. If you set this to 100 percent early reflections and then put in a pre‑delay of 90 to 100ms, you get the makings of a good doubling effect. Setting the reverb parameters for bright and short, with low diffusion and density settings, makes the doubled sound more distinct, and in this case I used only 14 percent wet, 76 percent dry to get the required balance. (On other platforms, any synthetic reverb plug‑in that lets you adjust the ER and tail balance should be able to create a similar effect.) This came close to the doubled sound Adam was looking for. One benefit of the Antares plug‑in is that the two voices it generates have slight and random pitch and timing variations and can be independently panned. Panning these hard left and right gave the main vocal a more believable doubled sound and added some dimensional interest to the mix.
The next vocal layer was treated with just a small amount of Logic's Overdrive plug‑in after being compressed, then Echo was added as an effect, with the delays sync'ed to sixteenths, and the effect level at around 25 percent. This chain was set up by Adam, so other than trading his distortion plug‑in for Logic's Overdrive and tweaking the settings, it ended up pretty much as he originally had it, but — as with other parts — less obviously distorted. Adam had laid down a further vocal part, using the Vocal Transformer in GarageBand to simulate a female backing line during the choruses. While not entirely convincing in isolation, this actually worked very well in context, so I kept it and just adjusted the level balance so that it sat under the other two vocal parts.
As I explained earlier, the bass line was played on a regular electric guitar, and then treated using the GarageBand Bass Amp plug‑in and EQ. I didn't change the basic processing chain much, but added a send to feed a Waves pitch‑shift plug‑in to add some sub-octave content, and also patched in a Waves Maserati Bass plug‑in (set to Synth mode) to shape the basic sound, making it more solid, and adding gentle modulation, to create what felt almost like a 'synth bass with pulse‑width modulation' effect: it would be impossible to replicate the exact sound of a bass guitar, so I thought we might as well make a feature of the bass sound. It was also necessary to roll off the high end from the sub signal above 150Hz or so, to add depth without it sounding like an obvious added octave effect, and the sub level had to be adjusted carefully, so that it just filled in that missing lower octave without being allowed to dominate.
For drums, Adam had mainly used Apple loops, augmented by the occasional sampled cymbal. One of the loops used towards the end of the song has something of a hip‑hop sound to it, and I enhanced this by using SPL's Transient Designer to shorten the decay time, imparting more of a beatbox effect. Apple Loops usually sound pretty good as they come, and this was no exception, so no other processing was necessary. All the drum parts were based on stereo loops that could be panned to the centre, as could the bass and the main vocal. Where two or more guitar parts were playing together, panning was used to separate them, while maintaining reasonable symmetry in the mix, and the same thing was done with the backing vocals. In combination with the rather more obvious stereo effects in the break section, this was enough to give the track all the width it needed.
As Adam was compiling his own album of songs, to include 'Deserts', I used Logic's Multipressor on the final mix, followed by Logic's Limiter, to create a sense of loudness and energy. Very low compression ratios were used in the four bands (less than 1.2:1), and the thresholds were set at ‑20dB so that most of the song's dynamic range was gently squeezed, rather than just the top few dBs being stamped on. The relative levels of the middle two bands were also dropped slightly, to create a hint of a smile curve, which gave the mix a bit more punch and sparkle. The limiter was set to catch only the occasional 2-3dB spike, so the mix didn't sound over‑processed.
I went back over the song with these processors in place and fine-tuned the levels, adding a little level automation on some of the vocal and guitar parts where necessary. Any added reverb was kept fairly short and bright to prevent it from 'washing out' the sound, and then I checked the overall balance from the next room, made some final tweaks and hit Bounce.
With a song like this one, you could continue making changes and refinements almost indefinitely, and some engineers might spend several hours fine-tuning the vocals, but I felt I'd reached the point where further changes 'wouldn't affect sales', and on the whole I was quite happy with the way the song had tightened up, with much stronger sounding vocals, less clutter and a generally lively vibe. Of course I didn't know at this point what Adam would make of my meddling!
Adam: "The first thing I noticed about Paul's mix was the introduction of the new 'evil' bass guitar sound, which immediately gave the track a heavier and darker feel — which is what I was after to begin with. It also meant that the main guitar could be lowered in level to allow the vocals and drums more space. Paul added a tempo‑locked tremolo to the guitar part, which gave a nice feel, and I was very pleased with that.
"I also noticed a dramatic improvement in the vocal sound, and the guitar‑amp hiss that had been obvious on the original mixes had gone. I felt there was a significant improvement in the overall sound, as in my mix the sounds had tended to blur into each other. Now the mix had more punch and definition.
"My original mixes lacked any sense of stereo width, as I wasn't sure about how best to use panning, and Paul has remedied this by panning my additional guitar and vocal parts. There isn't really anything I dislike about the new mix, and I think the new break section works really well, although I might have left more distortion on the voice where I sustain a long sung note over the instrumental section. However, the trick Paul did, stretching it so that it covered the entire section, was excellent — I loved it. In all, I absolutely loved the new mix of 'Deserts' and am grateful to Paul for explaining his mixing and mastering techniques to me.”
We've placed 'before and after' audio files on the SOS web site so that you can download them, audition them in your DAW, and hear for yourself the changes Paul made to Adam's mix. Go to /sos/jun09/articles/mixrescueaudio.htm for more details.