Mixing Essentials (Part 1)
Mixing in the MIDI Age
This two-part feature covers both the art and technology of mixing. The first article on page 48, 'Mixing Essentials', covers audio mixing in general, from a more basic level, and applies to any sort of studio mixing task. The second part, 'Mixing With MIDI’, starts on page 54 and covers techniques made possible by MIDI, specifically automated mixing and sending MIDI instruments direct to your master tape. Craig Anderton is your guide.
In theory, mixing should be simple: you take a bunch of sounds and adjust their levels and timbres until they're all properly balanced. Then again, playing keyboards should be simple too; all you need to do is press on the right keys at the right time. However, while anyone can bash on a set of keys and make noise, playing Clair de Lune is another story altogether. The same goes for mixing. Anyone can vary the levels of a bunch of different tracks, but making a mix that sounds good on a cheapo boom box as well as an audiophile's dream system is a much more exacting skill.
This two-part feature covers both the art and technology of mixing. The first article, covers audio mixing in general, from a more basic level, and applies to any sort of studio mixing task (mixing for live sound is a whole other topic, best left to its own article). The second part, 'Mixing With MIDI', covers techniques made possible by MIDI, specifically automated mixing and sending MIDI instruments direct to your master tape.
Craig Anderton explains what you need to know about the art, and the science, of making a great mix.
Mixing is what makes or breaks a tune. It doesn't matter how good your performance is if it's buried in the mix, or what kind of equipment you use if you can't distinguish one sound from another. Mixing is a creative act, just like songwriting or performing, and needs to be pursued with the same combination of enthusiasm, technique, and inspiration you'd apply to any creative endeavour. On a practical level, the mix is the last chance you have to present your music in the best way possible.
Fortunately, much of the mixing process is easily quantified and explained, so let's start with the basic procedures involved.
The four main elements involved in making a mix are:
1. Level-setting. This involves balancing the level of each sound with respect to the other sounds.
2. Equalisation (EQ). Carefully adjusting the timbre of each sound can make the difference between a muddy mix and a masterpiece.
3. Stereo placement. Stereo allows us to create a soundstage for our instruments. Proper stereo placement results in an open-sounding, full mix.
4. Ambience and special effects. Applying ambience, such as reverb and echo, adds a third dimension - depth - to the otherwise two-dimensional left/right soundstage created by stereo placement. Special effects include limiting, compression, phase shifting, pitch shifting, and other outboard goodies that add more character and/or animation to a sound.
What makes mixing so complex is that these four elements interact. Changing a sound's EQ also changes the level, changing the amount of ambience alters the soundstage, and so on. In a way, a mix is a 'combination lock' in which, once all four elements click into place, you have the perfect mix. We'll cover tips for each of these elements separately, but first let's talk about the goals for a good mix.
The ideal mix would do the following:
1. Allow each instrument, or sound, to be heard clearly and distinctly. Even with Phil Spector's famous 'wall of sound', you could pick out an individual tambourine or maracas part from a sea of drums, strings, and vocals. If the parts mush together, you don't have a good mix.
2. Be transportable to any system. Playback systems vary enormously. A well-mixed record will sound good on a bad system and exceptional on a good system. While it's possible to make a mix sound good on any one system, it requires real skill to make a mix so good that even a transistor radio with a 2-inch speaker can't destroy it.
3. Amaze, delight, and titillate the listener. Mixing offers as many creative possibilities as recording. Although you shouldn't leave too many problems to be 'fixed in the mix', the fact remains that a lot of songs have been saved in the mix, and ace producers and mixing engineers command high fees to improve already good tunes.
Now that we have our goals, let's figure out how to set our combination lock. First, though, a caveat. Different producers have different styles, and I can only tell you about my particular style. I also have some theories about mixing that are generally unverifiable and perhaps controversial, so remember that I'm just describing what has worked for me.
The most important tool is your ears, since the quality of a mix is directly proportional to how well you can discriminate between subtle differences in timbre and level. To improve your odds of creating a successful mix, listen critically to well-mixed recordings and analyse them. Strive for the same quality in your work as mixes by Bob Clearmountain, Shelly Yakus, Ron Nevison, and others of their calibre. Don't pay too much attention to the music; listen mostly to the mix. You'll sometimes find that bland pop albums are mixed to extremely high standards.
Although evaluating monitor speakers is a subjective topic beyond the scope of this article, I have some general recommendations about suitable monitoring levels.
The human ear is a marvellously delicate device. Exposing it to high sound pressure levels for extended periods of time can reduce, at the very least, midrange response (as confirmed by tests on recording engineers). This can ruin a career, because if you don't hear something correctly, you can't mix correctly. Furthermore, the ear has a sort of built-in natural limiting effect due to its logarithmic response, and monitoring at high levels will often make loud parts seem less out of balance than they really are.
When mixing, I prefer to monitor at extremely quiet levels for several reasons. First, a typical mix takes at least 12 to 15 hours, and ear fatigue can set in quickly if you don't monitor at low levels for most of the session. Second, I feel I can discriminate between tiny volume variations at low levels easier than at high volumes. Sometimes a lead line that sounds balanced at high volume will sound grossly overstated at low volume. Third, I've found that if I can make something sound satisfying at low levels, it will sound incredible when played back at high volume. If you have to crank up a mix for it to sound good, you haven't made a good mix - it's that simple.
I also try to get as many perspectives on a sound as possible. Rock musicians commonly will play a cassette dub from the master tape over a car hi-fi system as a 'reality test'; while this may sound almost superstitious, it's good practice to monitor a mix over a variety of speakers, headphones, and, yes, car stereo systems. This is why most studios have multiple monitoring speakers; no speaker system is flat, so by listening to a variety of systems you can take an average and mix accordingly. Also, no room is flat, so ideally you would not just listen to different speakers, but to different speakers in different rooms. Once you have a mix that sounds good on all of the above, your mission is accomplished.
Here are the steps involved in mixing a project. However, note that these often interact to a large extent, and at times you may find yourself going back to previous steps or anticipating future ones.
"If you have to crank up a mix for it to sound good, you haven't made a good mix - it's that simple."
1. Get mentally prepared for the mix. You are about to spend several hours doing something that demands total, unyielding concentration. Turn on the answering machine and turn off the phone's ringer, get a comfortable chair, adjust the lighting, and create a comfortable environment. Get out your track sheets for the tune you're about to mix and have a pencil and scrap paper available for taking notes.
2. Take an inventory of your tracks. Listen over speakers at low to moderate volume to refresh your memory about each track. As you listen, use masking tape or stick-on strips to identify which instruments appear on which mixer channels. Group instruments together logically; the rhythm section in one group of channels, vocals in another group, and so on.
3. Check for glitches. Nothing can kill a mix faster than getting bogged down in detours to fix little problems here and there, so get this step out of the way as early as possible. I generally clamp on earphones and listen to each track, in isolation, at a moderate level. Take note of when pops, clicks, out-of-tune notes, fret buzzes, and other problems occur. Either fix them or remember to do what you can in the mix to mask the problem. For example, with one old digital synth I use, the sound gets grainy and distorted at the end of its decay (due to the D/A convertors running out of resolution). In context with the rest of the instruments, you could hear there was a problem but the exact reason wasn't clear, until I heard the synth in isolation and noticed the distortion. The solution was to add a noise gate that muted the last part of the decay tail.
Headphones are best for this step and the next one, because they often reveal sounds in more detail than speakers.
4. If you're using 'virtual tracks' or mixing a tapeless (ie. all sequenced) tune, optimise each sound source's parameters. With virtual tracks, you sync a sequencer to tape and drive your instruments from the sequencer. The instrument outputs go directly through the mixing desk to the stereo master. Often you can optimise sounds at the synthesizer or sampler rather than at the mixing desk, which is recommended for two reasons: first, the less signal processing you use, the better (why boost the treble on an analogue synth when increasing the filter cutoff frequency often does the same thing?); and second, these 'tweaked' patches can be recalled instantly with System Exclusive data during the mixdown process.
5. Go back to listening over speakers and make a very rough mix of the 'naked' tracks. Don't add reverb, signal processing, or anything else to dress up the sound; we're still dealing with the basics at this stage. Listen for problems in the general gestalt of the mix. Are the vocals thin? Does the kick drum fight with the bass? Is the bass crisp and full, or muddy, or tinny? Do the synthesized strings melt like honey, or screech?
It's a temptation to slap on some signal processors at this point to improve individual sounds, but then you're working on the trees, not the forest. An important point about mixing is that any change you make to any instrument affects other instruments as well. For example, you might decide to add more reverb to the drums, only to find that if the strings are heavily reverberated, the two reverb sounds melt together and the strings sound less distinct.
Every action does have an equal and opposite reaction: change the bass, and you'll change its relationship to the kick drum; increase the rhythm guitar's midrange, and the vocals will have to be louder to be heard. This is why mixing is so very hard; even a couple of dozen tracks have a complex interaction with each other, and each needs to sound good on its own.
It's important to prioritise the order in which you'll work on individual sounds. I recommend optimising the most important parts first. In pop tunes, begin with the vocals first and make everything else subservient to the vocals. Drums and bass are usually next in importance. Work on the 'sweetening' - little percussive accents, strings, etc - only after the basics are together. You may want to start off your mix in mono, because placing sounds in stereo can eliminate some problems you might not catch otherwise. If parts fight with each other, you'll discover this a lot faster in mono. Besides, if you can hear each instrument clearly in a mono mix, you'll certainly be able to do so in stereo.
6. Start working on EQ. Proper EQ is the secret to getting a great mix. For example, if you want a bass to sound good over small speakers (which usually don't have good bass response), increase a bit of the upper midrange and treble of the bass track to give the sound more 'snap'. This helps the part cut through on little speakers, whereas a 'boomier' bass sound would get lost.
Again, remember that EQ does not exist in a vacuum. If you pump up the upper midrange for more intelligibility on a vocal track, this will tend to block out guitar, piano, and other midrange instruments. Think of EQ as something that makes all the tracks fit so that, when mixed together, they will fill up the entire audio spectrum without any one band of frequencies predominating. (A spectrum analyser is a great learning tool when mixing. I bought a 1-band domestic model, modified it for a faster response time by simply changing some capacitor values, and ended up with a very useful addition to my studio.)
As an example of how this works, consider a band with drums, rhythm guitar, lead guitar, piano, vocals, and string pads. With my voice, I find that a moderate upper midrange boost increases intelligibility and 'breathiness', thus making the vocals more human and distinct. To keep this from interfering with the background, I notch (cut) the rhythm guitar response a bit at the same frequency where I boosted the voice. (A parametric equaliser, which can specify centre frequency, boost or cut amount, and bandwidth, is best in this application, because you can home in on a particular frequency, make a deep notch - also known as a cut - with a narrow bandwidth, and effectively eliminate a few frequencies from the guitar sound.) The rhythm guitar then has a bright top and full bottom, but the midrange stays out of the way of the featured soloists, the vocal, and the lead guitar.
Drums have their own natural way of filling out the frequency spectrum, which is why they're often optimised first in a mix. (Another reason for optimising drums first, instead of vocals, is that with many tunes the rhythm is actually more important than the vocals or lyrics.) The kick drum handles the bottom, the toms the midrange, the snare the midrange and some highs, and the cymbals, the highs. By adjusting the level of each part of the drum set, you can do a lot of tailoring before you even have to touch the EQ.
In the bass versus kick dilemma (ie. how do you differentiate two similar sounds in a similar frequency range?), I usually select one as the primary 'boom' sound and the other as the 'thock' sound, depending on context. By this I mean that I'll make the kick a little boomier and give a mid or lower midrange boost to accent the initial beater hit, while making the bass a little less boomy but with an upper midrange boost to bring out the pick sound. Since the ear can easily differentiate between the kick drum beater and bass pick, it's easy to separate the two.
Why not just pan the kick to one side of the stereo field and the bass to the other? Well, aside from sounding really weird - we're used to hearing both sounds centred in the stereo image, and low frequencies are relatively non-directional - when cutting vinyl records, the low frequency content should be centred. The worst thing you can do to any record mastering engineer is split a bass part into stereo by putting one channel out of phase with the other; always keep low frequency stuff mixed to the centre.
Once you resolve the major EQ problems, get picky. If particular notes or narrow frequency ranges are out of balance, reach for a parametric equaliser. If there's an offensive frequency, simply boost the EQ on that track as far as it will go (making sure that the speakers are turned down, first, so that you don't blow up anything), sweep the frequency range until you dial in the obnoxious component, then cut (notch) that frequency by an appropriate amount to smooth things out.
An alternative to EQ is the 'exciter' type device. This increases the spaciousness of a sound not by adding treble, but by other means (eg. summing a controlled amount of distortion in with the original signal, adaptive EQ, or any one of several other processes). These units produce a brighter sound without necessarily adding stridency and are particularly effective on vocals, acoustic instruments, drum machines (for a little more crispness), and older samplers, which often tend to be a little lacking in the high frequency range.
Through proper EQ, each instrument can hold down its own part of the frequency spectrum without interfering with other instruments.
"Often you can optimise sounds at the synthesizer or sampler rather than at the mixing desk."
One final note: in a lot of cases, the job of keeping instruments from stepping on each other is more properly the function of a tune's arrangement. However, that's another subject, and in any event, it's often too late to make any substantial changes in a song by the time you're ready to mix.
7. Add any signal processing that is an essential part of a given sound. We're not talking about reverb here, but effects that actually change the nature of the part, such as a 16th note echo that makes an 8th note hi-hat part seem like a 16th note hi-hat part, or the phase shifting you put on a synth organ sound to simulate the effect of going through a rotating speaker. This still counts as part of the process of preparing the various tracks for a mix and should be done before you start setting levels and creating a soundstage - our next step.
8. Start setting up the stereo positions of the various instruments. Generally, the bassier components gravitate toward the centre of the stereo field, with sweetening and lesser parts living off to the side. Important, featured sounds should be recorded in stereo and spread across the stereo image. If a sound wasn't recorded in stereo, it can often be turned into stereo via judicious signal processing (see '49 Signal Processing Tips' article in SOS June 1989). Some single point-source sounds, such as vocals, often sound better with artificially created stereo than simply miking the voice in stereo. Drums are the reverse: miking them in stereo (or assigning drum machine sounds to stereo outputs) usually sounds better than trying to process them into being stereo.
Other candidates for stereo spreading are keyboard pads you want to use as a low-level 'wash', rhythm guitars, and any other parts that form a 'bed' for the lead instruments. Although the usual way to create stereo from mono is with the use of delay (split a signal in two, feed one to either the left or right channel, and delay one a bit before sending it to the other channel), as long as two signals sound different from one another and are panned to opposite sides of the stereo field, it will create some sort of stereo effect. In some cases I've equalised a signal with its bass components on the left and treble on the right, which doesn't create quite as strong an imaging effect as using delay but does spread out the sound a bit.
Any time you spread a mono signal into stereo, you run the risk of generating unnatural alterations should the signals ever be combined back into mono again (say, over an AM radio). For best results, always check these kinds of parts in mono to ferret out frequency cancellations, timbral imbalances, and other problems.
One additional point about panning is that I usually avoid panning a sound to the extreme left or right. It seems better to move the panpot at least a tiny bit toward centre from the extreme left or right positions.
9. Think about any changes in arrangement that will be made via mixing. One of the worst things you can say about a mix is that it's 'cluttered'. The mix is your last chance to remove parts that aren't necessary, as well as record any last minute emergency overdubs. Generally, less is more when it comes to mixing. The fewer the number of competing parts, the easier it is for the listener to focus on the most important parts of a tune. If you have a mono part that you've mutated into stereo, try having just one channel active at first and bringing in the second channel of the stereo pair when the song needs a bit of a boost. Always be conscious of anything you can do to increase the clarity, definition, and effectiveness of a tune.
With present technology, it's tempting to stuff a multitrack full of tracks on the assumption that any excess can be eliminated in the mix, but it's important to ask yourself whether a part adds to the song, or merely to the clutter. Then act accordingly. If you're absolutely sure that a part isn't needed, erase it from the tape (or sequence) so that you don't have to remember to push the mute button or pull down the fader when mixing.
10. Space: The Final Frontier. Now that the basics are covered, it's time to add the all-important third dimension of ambient space. Reverberation is a great effect, but be careful not to drown your tune in a sea of reverb. (If you've followed the tips above, your tune should already sound pretty good anyway and shouldn't need tons of reverb to cover up problems.) I pretty much find that two reverbs are a necessity: one to add general ambience to the overall mix (eg. the sound of a hall or room), the other to add special effects on specific instruments (such as gated reverb on a drum set snare, chorused reverb for guitar, etc).
Few budget reverbs have true stereo inputs, but they generally do have stereo outputs. This is important to consider if your mixer only has a single stereo auxiliary (or effects) bus and you want to use two reverb units. In many cases, you will get good results if you feed the reverb send signal into a single input and bring the stereo outputs back into the mixer as stereo returns, which means that you can then pan a signal to the left bus output to feed one reverb, or to the right bus output to feed a second reverb (Figure 1). And of course, you may want to feed both reverbs with one signal by panning the reverb send somewhere in between the left and right extremes.
Don't forget that additional signal processing can really perk up a reverb's sound. Running the reverb through a little bit of exciter or treble EQ can add sparkle; rolling off the treble a bit can simulate a more densely carpeted room with more absorptive surfaces. One time I even added chorusing to a reverb output as a special effect; while I wouldn't do this all the time, it worked very well in one specific case.
Back in the 1950s and early 1960s, the conventional wisdom was that you would add enough reverb to be noticeable, but not a lot more. With today's sophisticated digital processors, reverb is more than just a creator of space - it's often an effect in its own right that is integral to a tune's sound. Generally, though, excessive reverb should be added sparingly; use it on a particular snare drum hit, for example, or a lead guitar solo that you want to have sound like it's drifting up from the basement of a smoky club. Vocals generally suffer with excessive reverb, and one of the most common mistakes I hear on amateur tapes is drowning the lead vocal in reverb. It's better to add interest to the voice by doubling it, either by overdubbing a second part or via a short digital delay, and then add just a taste of reverb to fill it out.
In my opinion acoustic reverb still sounds best of all, but good luck finding a studio with a real echo chamber. You'll usually have to settle for a good digital reverb, but the advantage of digital reverb is that you can make it do a lot more than just simulate a room.
11. Go back and tweak whatever needs tweaking until everything sounds perfect. As mentioned at the beginning, all of these settings are interactive, and you'll probably find it necessary to go back and re-tweak sounds (making slight changes in EQ, amount of reverb, levels, and so on) until everything falls into place. If you hear something that irritates you, or something that isn't quite right, fix it now - or it will come back to haunt you every time you listen to the mix.
Mixing is a long, tedious process. I find it absolutely necessary to take a 2-minute break every couple of hours and get a snack, let my ears relax, answer some of the phone calls that the answering machine took while I was mixing, and so on. Far from disturbing one's concentration, a break lets you 'reset' your ears and restores a bit of perspective. You may even find that a part you felt was too loud (or soft) before you took the break now sounds just the opposite. When you're paying big bucks for time in a studio, it may seem that taking a break translates into losing money. In the long run, though, a bit of a break will usually save time and allow you to maintain whatever objectivity you have left concerning a mix.
12. Before you complete a mix, do as many 'reality checks' as possible. During much of the mixing process, I'll switch back and forth between headphones at low volume and speakers at moderate volume. But toward the end of the process, I'll play the tune back over hi-fi speakers, little Auratone monitors, the earphones on a Walkman, and so on. I'll even blast it once or twice! I'm also fortunate enough to have two sets of really bad hi-fi speakers, one of which has a mushy bass and the other, screechy treble. If the mix sounds good on all of these various transducers, I consider the mix a success.
If I'm working in a studio I haven't used before, I'll often take the first completed mix and play it back at a different studio, just to gain a different perspective. So far this hasn't uncovered any major problems, but just in case I'm ever at a studio where someone had come in before me and decided to play with the room equaliser, checking a mix at a different studio would be cheap insurance indeed against blowing a project (or forcing the mastering engineer to go through some real contortions to make your mix presentable).
The one point I can't stress enough is that, aside from mastering, the mix is the very last chance to make something sound good before it goes out into the world. Is it worth spending that extra hour to clean up a sound or enhance the vocal? Yes. Once a tune is mixed, it's etched in oxide.
One big luxury of mixing in a home studio, or booking a commercial studio for long stretches, is that you can leave a mix set up overnight and hear it fresh the next day. Usually I consider a mix finished at the end of the day, but I leave everything set up. When I come in the next day, I listen one more time to the mix and confirm that it is as good as I thought it was the night before. If so, fine; if not, I mix one more time with a couple more final tweaks.
Feature by Craig Anderton