Learn Music and Audio Production | iZotope Tips and Tutorials

7 Tips for Using Subtractive EQ When Mixing Audio

by Nick Messitte, iZotope Contributor August 10, 2018

Balance your own mix:

Neutron 3

iZotope email subscribe

Never Miss an Article!

Sign up for our newsletter and get tutorials and tips delivered to your inbox. 

This article references previous versions of Neutron. Learn about the latest Neutron and its powerful new features like Mix Assistant and Sculptor.

Subtractive EQ—the process of eking out unwanted frequencies from an audio source—is a powerful tool in the engineer’s arsenal. With just a few moves, we can make sounds appear less nasal, less harsh, less muddy, and more clear.

It’s such a powerful tool, as a matter of fact, that Nectar Elements includes a subtractive EQ slider. Upon analyzing the signal, a custom corrective EQ curve is applied, and you can tweak its implementation via a simple GUI.

For beginner producers and mix engineers, a tool like this gives you immediate results. You can really hear what well-balanced, subtractive EQ can do for vocals. It works great on a vocal—but you’ll still need your own, bespoke process if you want other elements to sit in the mix. A guitar might muddy up a certain range, or the snare drum might interfere.

And with that, I bring you seven tips for subtractive EQ, especially helpful for the beginner.  

1. Boost first, then cut

If you’re not sure which nasty frequencies need curtailing, go ahead and look for them with a big boost. Bring them up and out of the woodwork.

Now, the jury’s out on whether or not you should sweep a boost to scout for frequencies. Sweeping here means to move the boosted frequencies around the spectrum as you search for the unpleasant offender—but we’ll get into that later.

Whether you sweep or not, you can use the the built-in spectrum analyzer of many an EQ (I’m looking at you, Neutron 2) in conjunction with a boost to find what you need to cut.

Here’s a useful takeaway from this method: over time, you’ll train your ears to identify which frequencies to attenuate automatically. Once you do, you’ll find yourself foregoing this step.

2. Try a narrow Q

The “Q” is the width of a boost or cut—the number of frequencies you’re swooping in your grasp at any one time. As a general rule of thumb, try to use narrower Qs when cutting before going with the broad brush of a wider Q.

Why? Ideally, you’re trying to eke out only the offending frequencies. So, if it’s possible, you don’t want to inflict any collateral damage upon neighboring frequency bands, whose characteristics you might like.

When it comes to Q values, the higher they go, the narrower they are, so 20 is very narrow, while 1 is broad. Try something narrow first, and you can always broaden it later.

Neutron 2’s Masking Meter can help pinpoint frequency collisions:

3. Try a transparent EQ

Color EQs that add harmonics, pleasant smearing, and other goodies to the signal are great for boosting things we like. And yes, sometimes they’re also great for cutting.

But a lot of the time they aren’t. Maybe the EQ in question uses a proportional slope that won’t get narrow enough at low levels of attenuation. This means the Q is always changing in relation to the level of the cut, which may rob you of some control. Maybe the EQ sports a minimum-phase signature that smears this particular cut in an unpleasant, or does other unwarranted things to the sound.

Try something like Neutron 2’s EQ in digital mode, and you’ll find you won’t have this problem. What you’re cutting is only what you’re cutting—it’s very clean and precise in other words; indeed, this kind of EQ—called linear phase EQ—has often been described as “raising or lowering the fader on a group of frequencies.”

Yes, sometimes artifacts called “pre-ringing” can occur with linear phase EQ, and those with spectacular hearing do pick up on pre-ringing. However, a lot of the time this type of EQ still wins out when it comes to securing a naturalistic sound, especially when compared to a hardware emulation—unless the emulation you’re using is modeled on an inherently transparent EQ design. Then, of course, feel free to experiment!

4. Don’t overdo it on the level of a cut

There are times where it makes sense to cut out 3–4 dB (or more!) of an offending frequency—say, the ringing of a snare drum that’s too apparent upon compression. However, you’ll quickly notice that the more stridently you cut a range of frequencies, the less natural it’ll sound. It’ll often feel processes, thin, bereft of character, and chock full of unpleasant qualities now introduced by the very cut you’re using to ameliorate the situation.

5. Don’t overdo it on the number of cuts

When starting to curtail unwanted frequency blooms, we can often hack the whole track to bits if we’re not careful. The inclination is to tamp down the problematic frequency band, and then to look for other offenders immediately thereafter—sometimes we notice them straight away, sometimes they’re harmonics of the annoying bloom we’re trying to cut out in the first place. The result can wind up looking (and sounding) a lot like comb filtering, or just a hodgepodge of attacked frequencies that give off weird resonances and seem devoid of life.

So here’s a good modus operandi: make a cut, and then sit with it a while before going to the next one. Maybe it’s all you need!

6. Try a dynamic EQ when static doesn’t work

Sometimes, no matter what you try, a static cut at a given frequency just leaves the goings-on feeling thin and lifeless. That’s often because the ear wants to hear some of the frequency at first—a hint that it exists—before the frequency swells into offensive territory.

With a dynamic EQ, you can give the ear exactly what it wants before battening down the hatches on the nasty, upcoming frequency storm. With a medium attack time and a release tailored to the timing of the track, you can let through a little bit of that undesirable sibilance, or some of that 3 kHz harshness, or just enough of the nasal noise. A little bit of these otherwise harsh ranges do make the vocals sound like they are, in fact, coming from a human, or from a real instrument. So experiment with dynamic EQ when static doesn’t work.

With Neutron 2’s EQ, you actually have the option to switch individual bands in and out of Dynamic Mode, which means you can handle both static and dynamic issues in one plug-in. Plus, you can sidechain any frequency band to any other frequency band within the plug-in. This can come in handy, as with the following example:

Say you have a guitar part that’s chugging along, but it’s not meaty enough, and the pick attack is too annoying. You can set up a dynamic EQ to clamp down on that pick-attack, letting just enough of it through to establish the necessary attack. Then, you can set up a dynamic boost on the meat of the guitar, and sidechain it’s operation to the pick-attack attenuation.

Now, every time the compressor clamps down on the pick-attack, it’ll give a nice little lift to the warm frequencies of the guitar. This is just one of the many ways you can orchestrate dynamic, subtractive EQ within a plug-in. The use of boosts in this technique also segues us nicely into the next tip:

7. Use boosts to shape cuts

This one took me a long time to figure out, but once a fellow engineer showed me the technique, it completely changed how I employ subtractive EQ. Basically, you’re a bit stuck with the filter shapes you have in any given equalizer—a high-/low-pass filter, a shelf, a parametric notch, or a butterworth filter will always behave in accordance with its internal architecture. This might not always serve the material at hand.

If you find that the slope of a cut is curtailing frequencies in a manner that leaves the material sounding dull, this is where you can use corresponding boosts to shape the curve of the cut.

It’s best if I show this in a screenshot. Here’s a vocal being high-passed to 170 Hz, to get rid of unnecessary low-end rumble.

Vocal high-passed at 170 Hz

However, see the slope at the bottom the bottom of this notch? It’s the gentlest one we have, but it’s cutting things a little too close; we’re losing some lovely frequencies right around 160 Hz, where it could use more life.

Enter the boost to enliven the proceedings, like this:

Vocal high-passed at 170 Hz, add boost

Now we’ve shaped the high-pass filter beyond what the curve itself could do to better suit the material. You can use this technique with every kind of filter shape.

To sweep or not to sweep?

This can be a bit of a sore subject amongst engineers. Some don’t like to sweep their frequencies, feeling that it can color what you hear because you’re auditioning hundreds of “wrong” frequencies for that crucial “right” band. Some engineers feel they’re not affected by the process, and that the speediness of sweeping far outweighs any adverse side effects.

If you don’t want to sweep, you can set up the boost or cut in bypass and then switch it in. You’ll immediately know if it’s wrong, and if it is, disengage to bypass again and change it. Then repeat the process.

This takes longer, but if you truly believe you’re being affected by sweeping, it’s a good practice to cultivate.


In my nascent years, I spent a lot of time employing subtractive EQ to a judicious degree. Too judicious, I’d wager now. The truth, I’ve learned, is that a subtle hand can often get the job done, especially if transparency is my end goal. Like many other things in life, I was too zealous in my earlier years, and my maturation as a mixer has a lot to do with allowing myself do more with less.

So of all the tips I’ve outlined above, I’d like to reiterate the one about taking a moment after enacting each cut to sit, and to listen. Patience goes a long way toward improvement, at least for me. You can try to rush forward with as many cuts as you can slice, but if you’re like me, you’ll find that is a truly subtractive process indeed—and not in the way that helps.

iZotope Logo
iZotope Logo

We make innovative audio products that inspire and enable people to be creative.