Mythbusters: Subtractive vs. Additive EQ

If you’ve been out and about on the internet then you’ve probably read that subtractive EQ is generally better than additive EQ. In other words, it’s better to attenuate than to boost.

And you have probably read all sorts of reasons for this: less phase shift, sounds smoother, more accurate, doesn’t require active gain, etc.

Well, I’m here to tell you that a great deal of that is false.

Let’s start by dispelling a few myths.

Myth #1: Subtractive EQ incurs less phase shift, or less noticeable phase shift

This is untrue. The degree of phase shift is 100% controlled by the amount of cut or boost. In fact “phase shift” is almost synonymous with “equalization.” They are inherently part of the same process.

Now, are the artifacts of said phase shift less noticeable? In objective technical reasoning, no, they are not.

And here’s why: If you exclusively use boosting to create an EQ curve on a source and then recreate the exact same curve using only attenuation, and then level match those two outputs exactly, you will get identical signals.

A simple way to test this is by taking any sound source and a shelf filter. Mult the sound source. On one, use a hi-shelf to boost up some high end. On the other, use a low shelf, set to the same corner frequency with the same slope and attenuated by the same amount. Then boost the output volume of the low-shelfed signal up to match the hi-shelfed signal and flip the phase. They will completely null.

What’s important, is to use an EQ where the parameters match up equally. For example, this will work with the Waves Q EQ, but will not work with the Waves Ren EQ.

What this demonstrates is that the phase shift and artifacts produced by that shift are based on the curve and degree of equalization, not whether the equalization is positive or negative.

Myth #2: Subtractive EQ Sounds Smoother

Ultimately the truth to this is based more on application than reality. It tends to be easier to mix additively¬†— boosting up things you want more of.

The problem with this is that it leads to a lot of compensational boosting. By that I mean boosting up lots of frequency ranges when really we just wanted to hear less of one frequency range. Or we will boost up a frequency because we aren’t hearing enough of it, when in reality there’s something from another instrument that’s getting in the way.

These issues aren’t inherent to additive or subtractive EQ, rather, they stem from using additive EQ when subtractive EQ would be a better approach.

Myth #3: Additive EQ is Less Accurate

This is something I hear a lot, and quite honestly I’m not even sure what it really means.

I believe this myth stems from the idea that it’s better to “cut narrow and boost wide.” In reality you should boost or cut as narrow or wide as needed.

I think it’s easier to over extend the width of an EQ boost because that means more stuff is getting louder. And louder has that instant gratification effect of sounding better. Again, this is a result of application and not inherent to how an EQ works.

Myth #4: Subtractive EQ Doesn’t Require Active Gain

This one is based on the idea that because you are “taking something away” you don’t need to add power to do it. This is really only true in the simplest filter setups.

While it’s true that a simple resistor-capacitor circuit doesn’t need any power, most more complex circuits do. That’s not to say there aren’t complex subtractive circuits that work completely passively — there are¬†a lot. But many EQs, no matter what you are doing, are drawing on active power. It completely depends on the EQ itself.

Lastly, digital EQs don’t use power. There’s no such thing as an “active” digital EQ, outside of plugins that emulate hardware.

Myth #5: Subtractive EQ Frees Up Headroom

This is half true. Subtractive EQ can and often does reduce overall amplitude, but the process is actually not amplitude based. You’re not exactly adjust the level of frequencies – you are adjusting the phase, and then getting a change in level at the band you are adjusting. But that’s not the same as adjusting the signal amplitude. In fact, on occasion you will use subtractive EQ and the result will be a boost to the signal amplitude! I kid you not.

So what’s the difference?

The major difference in subtractive vs. additive EQ is the thought process.

When you are using subtractive EQ, you are thinking “what can I take away to make this better.” Taking something away lends itself toward degrading a sound — so any time you take something away and it improves the sound you pretty much know you are making a good EQ move.

Additive EQ, conversely, is enhancing a signal, which naturally tends to sound better no matter what (particularly if you are doing it in solo mode). The real key to using subtractive or additive EQ is to have a good idea of what you’re trying to accomplish. The “solo” button can be very useful here.

Lets say you want to hear more upper mids in your guitar lead. Solo the guitar — does it sound like it has enough upper mids? If not, then use some additive EQ and boost those mids. If it does sound like it has enough upper midrange in solo, maybe there’s something else in the mix that has too much upper mids like a piano or acoustic guitar. Or maybe you really just want to hear the guitar louder, but when you turn it up the lower mids become overpowering. In these cases, subtractive EQ is going to be more effective.

I hope that clears up the reality of EQ. I have another article that dives deep into equalization here.

Matthew Weiss

Matthew Weiss

Matthew Weiss engineers from his private facility in Philadelphia, PA. Credits include Snoop Dogg, Gorilla Zoe, Arrested Development, Dizzee Rascal, Gift of Gab, J-Son and many others. Get in touch at Weiss-Sound.com.
  • Tim

    Take that all you subtractive EQ snobs!

  • randyhoexter

    Nice article. Indeed, EQ phase shifts are symmetrical; boosting or cutting produces the same amount.

    However, high-frequency boosting makes these phase issues much more audible. This is because our ears are much more sensitive to phase information in this area. Most of our localization and image information is above 5 kHz. T

    The phase shift used when reducing frequencies is subjectively less noticeable. Listening tests show humans are much more aware of volume increases than decreases; in other words, we are better at detecting presence than absence…

    In addition, EQ is often employed to make the signal brighter, and grabbing the high-boost control first guarantees a spitting, harsh signal. This is why cutting other frequencies makes sense in this case, because subtracting mids will have less of a subjective effect. the phase shifts are lower.

    This means additive EQ is quite effective for lows and midrange. But watch those upper mids and highs…

    • Matthew Weiss

      Right on. A capture with a really nice top end you can generally get away with quite a bit of treble boost. Anything that’s already got some spittiness is just going to get worse. Either way I find smoother results often come from opening up the midrange first and then doing gentler boosts in the upper mids/treble where needed.

      It’s hard to say if this is the reaction to the EQ’s artifacts though. It may be compounded with improper use of the EQ. For example, a vocal may need a bit more presence – so a wide boost in the upper mids might be good. But there may also be a narrower peaky-harshness in that range that needs to get tamed down, and failing to address that will give you harsh results.

    • randyhoexter

      Those are good points, too… the narrower the bandwidth, the worse the side-effects. And vocals are certainly often handled poorly. Broadband filters are much better, I agree.

      However, we did a lot of tests with HF boosts using FIR linear-phase eq versus standard types with the exact same settings, and the result was much harsher in everyone’s perception. Of course, this is just a subjective test, but my feeling is that HF phase issues are the cause of this harshness, not inherent harshness in the signal.

      In general, I find that if you want to do a lot of boosting, linear-phase EQ is the way to go… like on cymbals.

      Conversely, if you are doing narrow-band work, the traditional eq works better because all the ringing is after the transient, whereas linear-phase introduces pre-ringing, or signals before the transient…which is very strange-sounding!

      Enjoyed your writing!
      all the best,
      Randy

    • Matthew Weiss

      These things get sticky. Your results may have something to do with the individual EQ designs. The ripple, ring, and resonances will still be present in an FIR design (vs an IIR). The differences are that the ring appears before the output of the signal – as you mentioned – which is definitely odd, and there’s no group-time delay. The “linear” part of the EQ refers to the phase-time response of the entire signal.

      For example, the Waves Linear EQ has a “low ripple” setting that mathematically adjusts the band width and shape in order to evenly distributed the fluctuation in frequency output. It sounds very smooth, and can’t really be replicated easily with a minimum phase EQ. In many cases I actually prefer this for things where transparency is a must. The drawback is that the group-time offset is how phase naturally works, and when compensated for by a broad time offset the result is a lot more two dimensional (although that is subject, but to my ear and according to the laws of physics).

      I find the best spot for linear phase is between phase-coherent material. Cymbals as you pointed out – in a drum kit, because each band will stay in phase with the rest of the kit.

      I wish there were more conversations like this in the comment sections of my articles…

  • ChrisPorro

    that’s a very good one.

    years ago i took a pro
    tools certification class and the instructor told us he would never use
    cheap eq plugins to boost a signal only to cut one. the thinking being
    subtractive eq is less damaging so you don’t need a sony oxford for it.

    i believed him. why not? he had been doing it a long time.

    then
    a few years ago i read a book… “mixing audio” i believe. lots of good
    technical stuff in there. the author made the argument you made. it was
    somewhat technical and when i read it i just knew he was right.

    fletcher-munson is worth mentioning since an eq boost will make the signal louder. those guys knew “louder is better!!!” i try and match levels and do a bypass to see if i like it better.

  • Glen Himmaugh

    As an Old school audio guy, 40 yrs plus, I noticed that the quality of the device being used was never mentioned, here we are in 2013 and the market, clubs, bands and even sound companies still have Music store quality devices in the audio chain, when faced with older Yamaha type products where noticeable noise is added with almost every move of the knobs whether it be an SPX90 or a 1/3 octave EQ we are almost better off without the signal processor at all,but when forced to correct the relationship between the room, the mic, and the speakers cut is usually the choice of most of us, the Use of an old school analog EQ (regardless of the brand) as a tonal control for an entire system at least in the old days would be frowned upon and thought of as dead wrong, instead of say a frequency correction IE tightening up the low section above the subs or the mids, pulling the harshness out of the compression drivers ……That was The rule in Recording studios as well, Flatten out the monitors by pulling & not adding (If you can help it) This made for a much quieter environment from the start…Even medium grade comp/limiters/gates and such had limited band with so as to make the device cost effective to musicians, so compression your Kick drum would add hissy noise and destroy the overall bandwith of the instrument so it was tight, but no crisp attack to go with the tight bottom end ……I know we live in a digital world know, but trying to work with the items that are still in the FOH racks today Less is Always More………But I’m old …Nice article …..Glen Himmaugh…. Island America Music/ Sound Of Key West

  • Ramilas Mamedov

    the benefit of subtractive EQ relies in psychoacoustics. By cutting you are not biased that much by perceived loudness.

Recommended