There has been a lot of discussion on Multisound about how to get a good mix. There are two parts of getting a good mix for me the initial mixdown, and the mastering step. I’ve mostly had to figure this stuff out for myself, because I’ve never found a good concise discussion of how to go about it. This first post is on mixing down.
EQ AND LEVELS
The first step to understanding how to mix well has nothing to do with mixing it has to do with listening, and training your ears. To this end, I advise spending some quality time with CDs that are in the style you’re aiming for. A good thing to have while doing this is a graphic EQ with metering for each band. Play with the graphic EQ, notching out bands to see how it changes the sound, and whether you can hide or accentuate individual instruments in the mix. You want to reach the point where you can pick out the characteristic frequency bands for each sound.
Keep your reference CD’s handy, so you can A/B them with your mix in progress. Often this is enough to guide you when you’re getting stuck.
The next step is really getting to know the characteristic frequency ranges of individual instruments in the mix. This is of supreme importance because of a pervasive audio phenomenon called masking. Masking is, in terms of psycho- acoustics, the phenomenon where a loudest sound in a particular frequency band is going to mask any softer sounds.
Why is masking important? Because if one sound masks another, the masked sound can’t be heard, yet it contributes to the overall amplitude of the mix. Since you only have a certain dynamic range to work with, sounds that you can’t really hear waste dynamic range. What’s worse a masked signal interacts with the louder signal, muddying up the signal you CAN hear, without contributing anything useful to the mix.
This leads to the first principle of mixing: Give each instrument a distinct frequency range in the mix. This has two results: 1) It allows every signal in the mix to be heard distinctly at a lower level. 2) It prevents different signals from interacting.
Applying EQ
Having a mixer with good parametric EQ is essential to applying this first principle. There is really no substitute for a good analogue mixer for this application. You can do it on the computer, but knobz make it MUCH easier to experiment, without time lag introduced by working with a computer. The steps I use for this are these:
Step 1
Solo the sound, and use the high and low shelves on the EQ to cut energy on each end of the spectrum. Start with bass (unless the signal in question is a bass signal) dialing down the low shelf until you notice the sound quality changing. Then raise the low shelf until you find the point where adding more bass becomes noticable. The sweet spot is about halfway between those two points.
You’ll notice that some sounds are affected very little by cutting bass, because they have no bass energy to speak of. Fine — don’t be afraid to jack the low shelf down all the way in that case.
Do the same thing with the high shelf — cut until it sounds different, then dial it up until it gets noticably louder on the high end. Then pick a midway setting.
Step 2
Use a sweep mid band, and jack up the Q (i.e. narrow the bandwidth) for that band almost all the way up. Sweep the frequency until you find the point at which that signal is loudest. This is the characteristic frequency for this signal. Back off on the Q to zero, and then notch the signal up just enough that you can see a couple of decibels of peak level gain.
Step 3
For signals that are primarily bass or treble (i.e. bass guitar and high-hat cymbals, for example) the process is similar, but you can be more agressive about trimming the far end of the frequency spectrum.
Once you’ve done this for every channel in the mix, you’re ready to listen to the mix with all instruments present. If you’ve done things right, the mix should already be a lot tighter than with no EQ engaged.
Now let’s take a break to look at a little decibel math. A signal that is -6db quieter than another is half as loud. If you add two -6db signals, their peaks, if they coincide, will be at 0db — i.e. as loud as you can go. Therefore you can’t really have everything at 0db without distorting. So set all the active channels to a fairly low level — -24db is a good place to start.
Now listen without touching any faders to the track and make some notes on how the mix works. Which signals stick out in the mix? which ones seem buried? What is the overall sound like? This points you to how to apply some simple rules:
- If a signal sticks out, lower it’s level in small steps until it fits in the mix.
- If a signal sounds buried, raise it’s level until it sticks out, and then apply rule one.
- If the mix sounds boomy in the bass, pull back the levels on all instruments that are bassy.
- If the mix sounds harsh on the high end, pull back the levels on all instruments that are characteristically trebly.
- If the mix sounds ‘midrangey’ (boxy, nasal, harsh) pull back the signals that are characteristically midrangey.
These rules need to be applied a little bit at a time, because you’ll need to iterate through the process a few times to get the sound you want. If you do to much at any one step, you’ll obscure the effects of all the previous adjustments. You will get the best results if at every stage of the game you change things only as much as you have to notice a change in the sound. With practice, you’ll get to the point where you can hear very small changes, and will generally do less tweaking than you do the first few times through the process.
The stingier you are with boosts and cuts the better, because your goal is to come up with a mix that best shows off all the sounds in the mix. At the same time, the sound you’re going for is smooth and even. If a casual listener can’t hear what you’ve done to the mix, then you’ve done your job, because it will sound ‘better’ without there being any one thing the listener can point to. You want people to hear the music, not the tweaks you’ve done to the mix.
Compression
Compression, while essential, is in my opinion frequently overused and abused. Like anything else, a light touch is better, especially if you’re using the sort of cheap devices that mere mortals can afford.
I for one tend to compress an absolute minimum, until the mastering stage, which is a separate issue. Where it is useful is in taming signals that have a wide dynamic range. Any instrument that is struck or plucked is a prime candidate for compression. This includes drums, guitars, and pianos.
The easiest way to do compression is visually. Most compressors have metering to help you out here. Your ears aren’t really very helpful, because the best compression is largely inaudible. I generally start with medium attack and release settings, and fairly low ratios. Watch the gain reduction meters and adjust the threshold to the point where the compressor engages only during peaks. You want to tame those peaks without running the compressor all the time, because unless your compressor costs as much as a good used car, the less it is in effect the better.
Ideally you will have a compressor on every track, because if you tame the peaks in the signal, you can raise it’s level without overloading the mix. Compressing before equalizing can also give you some headroom to work with when doing EQ — since changing EQ changes signal volume, you can do more with EQ before overloading if the dynamic range of the signal is reduced beforehand.
Compression can also be used to make sounds punchier. A short, but noticable
attack setting lets the peaks through while lowering the overall level of the signal. This is especially useful with drums, since it emphasizes the attack, making the sound stick out in the mix, while lowering the average volume of the signal.
In general though the rule of thumb is this: get some gain reduction on the peaks, but keep the compressor wide open for most of the time.
Overall Advice
When I started out messing around with making tracks, a lot of the stuff I’ve described above seemed really mysterious and hard to grasp. The breakthrough for me was realizing that all the big-bucks pros were doing was USING THEIR EARS IN ORDER TO MAKE MUSIC THAT SOUNDED GOOD. No more, no less. The question you always need to ask yourself is, am I listening to what I’m doing? This may seem rather obvious, but it’s all to easy to get your knickers in a twist over technical details, but you’re going to do a whole lot better if you just LISTEN. Focus on one thing at a time using rules of thumb (as detailed above) to get the process rolling. Then, let your eyes go out of focus and just listen for things that bug you in the mix. If you hear something that bothers you go back again and listen until you can zero in on what it is that cause the problem. Then try and fix it with the least amount of modification to the mix.
What is the overall theme of the method I’ve described? This: You work on a mix best when you’re subtracting, not when you’re adding. The trick is to know what to take out!
Download this tutorial as a PDF file
