Pro mixers do great work because they're not distracted by problem-solving. With the right preparation, you can do the same.
How much of your last mix session was spent cleaning up tracks and fixing errors, rather than actually mixing? Dealing with a long list of problems can take hours and tire your ears, when you should be sprinkling fairy-dust and weaving magic! Top mix engineers interviewed in SOS and elsewhere invariably claim to prefer to 'work fast', and that means working fresh, when one has the energy and mental clarity to be open, inspired and objective. The opposite happens when a mix drags on into the small hours: no matter how supreme your levels of concentration and stamina, fatigue kicks in, ears get tired, and the lines between reality and illusion becomeincreasingly blurred. We've all been there.
The ability to mix quickly largely comes with experience, allied to a certain amount of skill — and in the cases of the world's top mixers, the work of assistants whose job it is to organise and clean up individual tracks, making them truly ready for 'just' mixing. The following tips aim to help the typical home/project-studio producer reach a similar situation, whether completing their own recordings or mixing tracks for a client.
Working in one's own studio has obvious time and cost benefits that we all appreciate, but the endless freedom available can lead to indecision, with whole collections of songs becoming permanent works in progress. Committing to a good old-fashioned final mix session, especially one involving the outlay of hard-earned cash in a commercial studio, creates a deadline, which forces a track's completion and allows you to move on to the next song.
In order to maximise the chances of a successful final mix, you need to do a premix: a preparatory session, or sessions, used to tackle any issues in digitally recorded tracks, and to do a little housekeeping. This may require some time and energy, but splitting the work into distinct premix and mix modules makes it far more likely that you'll do the best possible job. Once the preparation is done, you should be able to start a mix with everything ready to go — just like a pro. Getting the spadework done beforehand means mixing faster, fresher and better.
Identify the song's strengths and weaknesses. If it's your own composition, try to be as objective as possible. That can be difficult, so enlist outside help if necessary. Getting a trusted friend's opinion before you start can be invaluable, as all the mixing in the world won't help if the song has serious flaws in its construction, such as the intro being 10 minutes too long or the chorus being even more boring than the verse. The ideal position to be in is one where you are totally happy with the content you're about to mix: the tune should sound decent enough in its rough state, with your final sonic tweaking merely adding a fine icing to an already delicious cake.
Of course, if you're mixing for a client, especially an experienced artist who knows what they're doing, hacking into the song's structure may well be beyond your remit, in which case you should just get on with the job. On the other hand, if your client is inexperienced and obviously needs help, you can and perhaps should make suggestions, but make sure you do so with care and sensitivity; forcibly imposing your 'vision' can quickly lead to a breakdown in relations! Diplomacy, especially with bands, is essential: listen to and respect the ideas and wishes of all concerned (even the drummer). Once you do this, you're more likely to encounter a positive reaction to your own ideas, which you can then implement with confidence, rather than having to second-guess and wonder "Are they going to freak out if I do this?”
Happily, if you're trusted enough to be hired to mix a track, clients will often give you completely free rein, in the hope that you'll come up with something wonderful that they'd never thought of. As a creative mixer, this is a great position to be in; just make sure you really are in that position before you wield the scalpel!
When working on a self-recorded track, it may be tempting to go straight from tracking into mixing mode, but it's worthwhile at this stage to copy the session and all its associated files to a new location, and then proceed from the copy. The best option is to copy it to a different drive, as this creates a backup that will get you out of trouble in the event of a disk failure (which will happen to us all at some stage). And while copying to the same drive doesn't provide this safety net, it does at least give you the freedom to process and 'hard' edit files with the knowledge that the originals are intact elsewhere, should you go too far.
Having separate 'originals' and 'mix' folders is also useful if the project is ever to be sent on to another engineer, allowing you to choose whether to send only the recording session files, or the results of your mixing labours.
At this point it's probably worth taking a fresh look at the DAW's arrange page. Many sessions start cleanly enough, with drums, bass and guitars sitting neatly atop, but descend into a mass of solos, backing vocal takes and random overdubs towards the bottom. Tidying this up now will aid navigation later, so group recorded elements logically, with similar instruments on adjacent tracks. Colour-coding mix elements is also handy, though whether you use it to differentiate between instruments or between sections of a song is a matter of personal choice.
A typical mix will usually need subgroups for guitars, keys, backing vocals and so on. While linking channel faders is an option, using a good old-fashioned bus arrangement has the advantage of allowing overall audio processing of each group, and makes it easy to create stems later on. In addition, you can adjust relative levels without having to 'de-clutch' the group. Set these up now, and leave each group's fader at unity until its internal elements are at least roughly balanced.
If you're importing a bunch of files into a new mix session, it makes sense to bring them into a pre-made mixing template. Having such a template saves you from reinventing the wheel; why spend time recreating your 'standard' channel setups, subgroups and effects chains every time you set up a mix? Forget the common misconception that using templates will make everything sound the same. They don't! The template is just a starting point, from which you will almost inevitably digress as a mix progresses.
Making the template can actually be good fun too, as it gives you the chance to create your ideal mixing desk. Each channel can be pre-armed with favourite plug-ins, with various sends routed to go-to reverbs, delays, wideners, parallel compressors and so on. If your DAW doesn't already feature a gain knob at the top of each channel, start each chain of inserts with a simple gain plug set to -15dB, to prevent overloading channel plugs and the stereo mix bus (unless your audio has been recorded at a particularly low level, or you prefer to work with clip volumes in the arrange page, in which case you just bypass the plug). The -15 trick also helps keep channel faders close to their 'unity gain' setting, where in most DAWs they act with optimum precision.
If you're working in the side-chain-pumped worlds of dance or pop, having a permanently set-up bus compressor that's triggered by a drum machine will be labour-saving. Finally, using a uniform set of processors across the mix bus — typically EQ, compressor and limiter — can help provide some consistency between mixes, even if they're performed weeks or months apart (you can always bypass them if they're not required in any given mix).
Most DAW software features some form of 'take folders' which facilitate, and even encourage, the recording of multiple performances — typically of vocals and tricky instrumental parts — the best bits of which can later be identified and 'compiled' to produce a perfect track. The actual compiling is often, by necessity, put on the back burner at a recording session ("we can sort those out later!”). A singer, for example, having completed that day's lead vocal duties, may prefer to immediately track backing vocals, or go to the pub, rather than help sift through a mass of takes. And as comping is a repetitive task that requires concentration, it's a task best performed at this premix stage, ideally by one or two people in a quiet room. Having a whole band offering endless opinions can make an already tough job impossible.
When comping a lead vocal, it's vital not to lose sight of the wood for the trees. The logical, methodical approach of going through each line, take by take, and choosing the best is not necessarily the most effective, or efficient: you can toil for hours and end up with a patchwork of individually 'perfect' parts each subtly tonally and emotionally different from the next, with the feel of the song ruined. It can be better, not to mention a lot quicker, to work in sections, deciding on the best overall take for a verse or chorus, and dipping into alternative takes only when a problem is encountered.
Once the comp has been established, bounce it down to its own file, for both ease of use and safety; in my experience at least, take folders can be susceptible to gremlins over a period of time. It can also be worth creating a double-track while in this vocal-comp mode, for use either throughout or in parts of the song (see my article 'Vox Pop: Crafting Perfect Pop Backing Vocals', in SOS January 2013: /sos/jan13/articles/vox-pop.htm).
Removing headphone spill, guitar hum and all the usual random coughs and mumblings from supposedly 'quiet' sections is a relatively easy but time-consuming task, so be prepared to get busy with the editing scissors. Ensure your DAW is set to seek zero-crossings when cutting, but do please be aware that this is not a cure-all. Once an edit has been made, check for clicks and pops at the region's start and end points, especially if emerging from or going to complete silence: applying a very quick fade-in/out to the region will usually cure any glitches. In a few more stubborn cases, a small adjustment to the start or end point may be needed.
Headphone spill, while not necessarily a problem in the middle of a rockin' tune, can be an issue during quieter moments. In bad cases, this simply cannot be remedied, in which case hiding the spill behind a hi-hat or shaker may be the only viable option.
Similarly, hum and buzz from electric guitars and old synths may not be hugely evident in busy sections, but the effects are cumulative. Thankfully, as chopping tracks into regions and muting the 'un-played' parts is a simple job, a lot can be done by eye as much as by ear. For a 'produced' sound, muting/un-muting regions bang on the bar can be effective (assuming they've been played well!), though it depends on the type of music: just as many parts will require, for example, a guitar chord to fade naturally into the next section. In those cases, edit at the point where the background noise becomes louder than the guitar, then apply a long, gentle fade to help the notes fade to silence.
Region editing is also great for tom tracks, which should almost always be cleaned, as the spill between strikes rarely contributes anything to a kit's sound other than a resonant woolliness and potential phase issues. As the vast majority of songs feature less than a dozen tom hits, it's usually a pretty quick job; you could spend much longer setting up a far less effective gate! Roughly speaking, edit the regions to allow the drum to ring up until the next loud hit (typically the snare), then shorten and/or apply a fade-out to taste.
Although crafting the vocal sound and its effects treatment are obviously major parts of the final mix, some of the basic chores can done in the premix. A useful psychological trick is to imagine you're about to send your comped vocal to your favourite producer, who most likely would not be impressed with a smorgasbord of rumble, coughs and spill. Of these problems, rumble is the most easily dealt with: a simple, steep high-pass filter, initially set around 100Hz, will usually do the job. Fine-tune the frequency to best eliminate low-end intrusions without affecting the vocal sound. Coughs, footsteps and other random noises are easily removed by trimming regions as discussed. Just take care not to chop the end of trailing words, and only clean up until the breath before the next phrase: always keep that breath!
Note, however, that the above assumes you're working with a typical pop/rock/dance track, with plenty of instrumentation surrounding the vocal. If your song is largely acoustic, and quite bare, reducing the level of noises rather than wiping them may be the better option, as an instantaneous drop to complete silence could be noticeable and jarring. Keeping a vocal track continuous and complete can also help provide some atmosphere and 'glue'.
Smaller noises, such as mouth clicks and plosives (the nasty thumps caused by blasts of air accompanying Ps and Bs), require forensic surgery. Editing the file itself is the best, and often only, option when dealing with such events, which may last for just fractions of a second. It's also quicker and more accurate than using arrange-page edits or automation. And it's more elegant, as your edits will still apply if the vocal phrase is repeated elsewhere in the song. Some producers shy away from hard editing, perhaps terrified by its finality. However, depending on your DAW preferences, most hard edits are actually undo-able to a certain degree, at least until you run out of undo levels or save the session, and if you followed my earlier advice, you'll have the original session to fall back on.
Using the sample editor, zoom in closely on the problematic audio, taking care to select only the offensive part and nothing else — the jagged spike should be easy to spot. Then apply a level drop of 3-4dB. Have a listen. Not enough cut? Repeat the process. Too much? Undo, then repeat with a smaller value. When an especially long air blast continues under the rest of the word, try using some low-frequency EQ cut, either in addition to, or instead of, the level drop.
Many users have reservations about de-essing processors: the examples bundled with many DAWs tend to be either ineffective or over-zealous (it can be extremely hard to find an ideal setting), while the better third-party offerings can be expensive. The good news is that doing the job by hand is not that difficult, and can produce results superior to those of any plug-in. It's curious that the practice is not more prevalent.
While it may seem an idea to tame sibilants using a high-frequency EQ cut, in practice a simple level drop often works best. After all, you'll usually want to keep the high frequencies that make up the 'S' sound, but just have them at a lower level. An EQ dip will do something entirely different, maintaining most of your selection's level while changing its sound, and can result in lispiness — you can end up with a loud 'Th' rather than the desired quiet 'S'.
Using a moderate level of zoom, sibilance shows up as dark, dense and easily selectable areas in a word's waveform: zoom in closer to help fine-tune the selection. Because of the brightness of many commonly used condenser microphones, and the ear's sensitivity to this frequency region, it may be necessary to impose seemingly brutal levels of gain reduction: try 6dB for starters, and be prepared to add another 3 or 4 on serious offenders. Be aware that your now-reduced esses will come back up in level somewhat during the mix, depending on how much compression is subsequently added to the vocal. If this is worrisome, set up your vocal track to play through a compressor while editing, in order to monitor and compensate for this effect. With a little practice, a complete vocal track can be dealt with in half an hour or so (assuming you have not made a song out of 'She Sells Sea Shells On The Seashore'). And because you're dealing directly with only the esses, the rest of the vocal remains 100 percent free from coloration. How many plug-ins can truly claim that?
Really obvious flaws such as timing and tuning problems need to be addressed, no matter what direction the mix takes. Remember, your aim is to make the subsequent mix session a pleasurable, creative experience, rather than a series of problem-solving exercises. This premix session is the problem-solving exercise, and by the end of it, you want to be able to knock out a static monitor mix that can represent the tune without embarrassment.
Timing and tuning issues warrant full features in themselves — and are thus beyond the scope of this article. (Mat Skidmore's Melodyne workshop in SOS July 2013 covers much of that ground, even if you don't use Melodyne: /sos/jul13/articles/melodyne-vox.htm). But I will say that with timing edits, make sure you correct any major, obvious mistakes before dealing with minor imperfections, and in general, aim to get things tight without killing the feel. If something sounds fine without lying bang on the grid, leave it well alone.
Similarly, serious tuning boo-boos should be corrected, but take care not to lobotomise your vocal by trying to make it 'perfect'. A pitch-correction plug set at a gentle value with a slow attack should be enough to smooth out the rough edges in a good vocal. If you're using something like Melodyne for more detailed,drastic or creative re-pitching, let your ears, and not your eyes, make the decisions. Once your changes are complete, keep your work safe — and your system stable — by immediately bouncing the results. Pitching programs are wonderful tools, but they don't seem to like being left to run indefinitely!
Place the tuned track adjacent to the (now muted/inactivated) original in the arrange page. This provides a handy escape route if you need to back-pedal later, and having both versions available to cut between during a mix can be a life-saver.
Recording, or printing, a musical part driven by a soft synth may reduce 'flexibility' — but you're now at the stage where decisions have to be made. Endless options are the enemy! If you've lived with a keyboard pad or synth line for weeks and have had no good reason to change it during that time, it may as well be printed. At worst, if the recorded sound really doesn't suit the mix as it progresses, you can re-run the part with a different sound. It's no biggie.
Generally, the pros of printing far outweigh the cons: your CPU will be freed up, often by massive amounts, for effects processing, and there'll be no more hanging notes and synth-crash headaches. Furthermore, many routine mixing operations are more easily accomplished with audio: for example, a fast fade-out to emphasise a transition can be applied to an audio region practically blindfolded, while achieving a similar result with a synth running live can require MIDI editing and/or automation. Finally, if and when you return to the song in six months' or five years' time, printed synth tracks obviate the need to track down an instrument or patch which may no longer exist on your system!
Although requiring a little more thought, guitars, keys and backing vocals which were multitracked for thickening purposes can also be rendered to audio in the form of a stereo bounce, for similar reasons: the bounced tracks will reduce screen clutter, lessen the strain on your DAW, and will also be much easier to edit.
Making bounces as useful and flexible as possible is key, so reverb or delays should not be added at this stage: you'll only get their levels exactly right by pure fluke. EQ treatments should be 'safe' and corrective rather than creatively extreme. With backing vocals, for example, get rid of any low-end muddiness with a filter, but leave the crucial mids and highs alone, because their exact tonal shaping is part of the mix job. Compression, on the other hand, can usually be safely added to either individual parts and/or the whole group, as BVs rarely need to be particularly dynamic. Also, if the parts are good and tight, the bounced BV is ripe for the manual de-essing routine mentioned above, which is a whole heap handier than working on each individual part.
As a rule, bounces should be made as 'wide' as possible, with things panned hard left and hard right, as this can be easily 'narrowed' at mixdown. Doing the opposite and trying to widen a narrow stereo bounce is rarely satisfactory, producing vague imaging and unwanted phasing artifacts, which are especially apparent in mono. However, premix bouncing does have its limits, and shouldn't be used in situations where the final stereo panorama is complex and hard to predefine (because of varying degrees of panning): it's best to leave those ones to the mix.
After a bounce, don't wipe the source tracks. Put them in a folder, or simply mute and hide them, for easy access should you need to return. You probably won't need them, but Sod's Law dictates that if you delete them you will.
Finally, bounce down a good rough mix in which you should concentrate solely on the balance between instruments and vocals without worrying about processing and effects. Import the rough, and place it (muted) in parallel with your tune, ready for quick comparison.
As well as providing a solid starting point, a good basic rough mix which highlights the energy and original intention of the song can be an invaluable reference during the later stages of the mix proper, especially if you suspect you've overcooked things with a wacky idea too many. Strident effects and bold edit cuts can sound great in themselves, but check that they really do work in context and add to the tune's momentum. Sometimes, howevergalling it may be, you have to admit that the rough — or perhaps just part of the rough — sounds better than the mix, in which case you'll have to back-track... but at least you'll know in which direction to head.
For a tonal reference, find a commercially released track in the same style as your song, and again, load it into the DAW in parallel with your song. Depending on how many styles your track encompasses, you might want to think about referencing short sections from more than one song. Remember to temporarily bypass any master bus plug-ins when checking either reference, as these will obviously affect the references' sound. Also, bear in mind that if you're referencing against a mastered track, you'll want to drag it down in level so that it's comparable in loudness with your mix in progress. By keeping your mix roughly within the treble/bass frequency extremes of the reference, you won't go too far wrong. Broad corrections can be made at the mastering stage, but it's infinitely preferable, and a whole lot more satisfying, to get it right first time!