What frequencies (hi and low) should be rolled-off to clean up drum loops?

I'm trying to cut back on harshness, mud, and artifacts for over a thousand drum loops. Any suggestions for hi and lo pass parameters that I can globally use with some confidence during batch processing? Thank you.

 

Hello @tommyzai, read the following:

High Pass & Low Pass Filters: Why You Should Master This Essential EQ Technique
https://isina.com/magazine/high_pass_low_pass_filters

How to EQ a Drum Loop --> https://aleksigodard.com/how-to-eq-a-drum-loop

 

Since you aren't making dance/bass music, you can probably come up even higher; but highpass at 30hz; and lowpass at 18khz. Those are being on the "safe side", and I usually have my DAW channel EQ default to that setting to load that way automatically. Batch Processing is a little different, because you cannot really put back the information you cut. I much prefer the default DAW channel EQ approach, because it is not applicable only to your 1000 drum loops. It's every channel you load that eq on, with room to adjust. Same time saver, but non-destructive.

 

it really depends on the frequencies of the drums but I usually roll of the bass at 50/60hz and give it a bump if its a hard deep kick, then roll the top down to 16khz...but say you want a persistent snare you would raise a band around 200z, BUT that's also where the worst mud usually is, so if the snare sounds fine then turn it down around 200hz. Then I just weave the eq around towards the middle high end until I hit the right frequencies I want or what blends best with the mix. I'm sure others have there own techniques, but there ARE some basic fundamentals to easily follow like that...first and foremost though is to just have a nice clean sample or better separate drum tracks!

 

He's talking about batch processing a thousand drum loops in the first post. It can't "depend on the frequencies", because when you batch process things you are running the same exact settings across every file within the batch. It does not mean you will not need to adjust your EQ settings on each one to taste or as needed; which is why I told him to not even bother batch processing them. You create a preset in your EQ of choice, or set your INIT/Default settings for your DAW's channel EQ to some conservative settings. That way your cuts are applied by just loading the channel EQ, and when you have them in your mix, you can eq further whatever way you want.

Batch processing 1000 files doesn't make sense unless you only want conservative settings applied to tweak later; or if you are creating a sample pack or library of samples from your own plugin or hardware synths and samplers. Otherwise it is quite possibly a waste of processing time. It's a holdover from old workflow where preprocessing samples was done on our slow computers to not have an eq on every channel. Like tops and tails days in your offline wave editor i.e around 2001 or earlier.

 

Are there some global roll-off settings for drums? For example, is there a need for anything under 30 or 40hz or above 18 or 20khz? I'm trying to batch remove any unnecessary frequencies, then manually adjust EQ as needed. My reasoning . . . this gets rid of any unexpected rumble, hiss, or artifacts.

 

EQ-Cheat-Sheet.pdf (3 MB)
https://workupload.com/file/vey3rEAP3db

That one has a sheet just for drums. Or, look at what Sonnox shows for the numbers even more easily, top and bottom bars describe each range:

 

Sorry to break it to you, but drums cover pretty much the entire frequency spectrum. Kick drums have your lows, snare and toms are in the mids, and hats and cymbals are in the highs. Essentially, there's no specific frequency you can cut/boost to make drums disappear, because they're at every frequency. That said, there are plugins dedicated to sound cleanup. Check out iZotope RX.

It's actually not possible without really ruining the other parts of the sample. It was best explained like this: You mix two colors of paint and then try to remove just one color. Basically, sounds share the same frequencies, so you'll never get it right.

 

I keep reading that hi and lo pass filter plugins cause artifacts and worse . . . phasing. Any truth to this? I don't want to use another plugin like the one from Voxengo to de-phase unless I have to.

 

Well firstly, you already know garbage plugins are capable of all kinds of nasty results; so here is a better way for me to answer the rest of the cases which are about 99% of the time anyway. I'm not due for a typing test anytime soon.

You:
true or false this forum post: "I keep reading that hi and lo pass filter plugins cause artifacts and worse . . . phasing. Any truth to this?"

Spoiler

ChatGPT said:

That statement is partly true, partly misleading. Here’s the breakdown:

• True in some cases:
  
  • Filters (hi-pass, lo-pass, band-pass) are usually implemented with IIR or FIR designs.
    
    
  • IIR filters (most EQ plugins, including simple HP/LP) do introduce phase shift by design. That’s not necessarily a “problem” — it’s just how analog-style filters work.
    
    
  • FIR (linear-phase) filters avoid phase shift but can cause pre-ringing artifacts and higher latency.
    
• False / Misleading:
  
  • Saying “they cause artifacts” as if it’s always bad is misleading. Properly designed HP/LP filters are extremely clean. Any “artifact” is usually phase shift (normal), or resonance at the cutoff if the filter is steep or underdamped.
    
    
  • In normal mixing, this is rarely an audible “problem” unless you stack many filters or use extreme settings.
    

So the forum post is not outright true. The reality is:

• Yes, filters alter phase or can create minor artifacts depending on type.
  
  
• No, they don’t inherently “worsen” your mix just because of that — it’s standard, expected behavior.
  

Do you want me to explain how to test your own filters so you can actually hear the phase/artifact differences?

You said:

yes

ChatGPT said:

Perfect Here’s a hands-on way you can test if a hi/lo-pass filter is causing phase shift or artifacts in your DAW (Logic, Cubase, whatever you’re on):

Test 1 – Null Test for Phase Shift

1. Take an audio track (drums, vocal, or a sine sweep is best).
   
   
2. Duplicate it to a second track.
   
   
3. On one copy, insert the high-pass or low-pass filter plugin you want to test.
   
   
4. Flip the phase/polarity (Ø) of one track.
   
   • If the filter was perfectly transparent, the two tracks would cancel (silence).
     
     
   • What you’ll actually hear is the phase difference introduced by the filter.
     

Result: you’ll hear comb-filtering type artifacts (hollow sound), showing how much phase shift the filter creates.

Test 2 – Sweep Test

1. Generate a sine sweep (20 Hz → 20 kHz) with a test tone plugin or sample.
   
   
2. Run it through the filter at different cutoff slopes (6, 12, 24 dB/oct).
   
   
3. Watch with a spectrum analyzer (FabFilter Pro-Q, SPAN, etc.).
   
   • Look for resonance peaks or ripples near the cutoff.
     
     
   • Any pre/post-ringing (on linear-phase filters) will show up as faint “echoes” around the transient edges.
     

Test 3 – Impulse Response

1. Feed the filter a single-sample impulse (like a click).
   
   
2. Record the output.
   
   • An IIR filter will smear the impulse forward in time (phase shift).
     
     
   • A FIR / linear-phase filter will smear forward and backward in time (pre-ringing).
     

⚡ By doing this you’ll hear whether the “artifacts” are real issues, or just expected behavior. Most of the time, you’ll find the “bad phasing” people complain about is just normal filter phase shift that your ear interprets as warmth or character.

Want me to make you a step-by-step Logic Pro version of this test, so you can run it right away with your current setup?

 

If you're worried about any possibility of having a shitty HPF/LPF, maybe consider getting the TDR Special Filters bundle. It's worth every cent, and it's very cheap most of the time. And you also get Arbiter that is the best de-esser ever made (and that's not even why you should be using it).

 

How do arrive at this statement and what support do you have available for it? Another null test and delta comparison?

Which others did you test it against? AI list does not include it or DMG's, but does include Softube's, Weiss, and Eiosis; with Sonible's Smart: option at the bottom of the list. Also, Fabfilter first and Sonnox 3rd, makes "opinion" based rankings likely.

 

Arbiter is not a de-esser, but can be used as a de-esser the same way you can do it with a multiband compressor or dynamic EQ. The way Arbiter works is why it is the best de-esser. Many de-essers are incredibly good (I use smart:deess or Velvet early in the chain), but none can fix high frequency sibilance like Arbiter does.

Instead of me trying to convince you that it's the best and listing all the dozens of de-essers I have tried in the past, you can just get the demo. No limitations, except for no recall. Maybe also check their own walkthrough video hosted by Dan Worrall to learn the controls.

 

I second that emotion. It's a damn good plugin which can also check plosives...and will cure your dandruff too!

 

Hello @tommyzai, prick up your ears and pay attention to all the nuances of the sound. Everything matters.

Set the monitor levels to a comfortable range—not too loud, not too quiet. If the volume is set too low, you might be misled by the non-linearity of the speakers and overcompensate for areas that don't need it. If you monitor too loudly, certain frequencies can be masked or overemphasized by the non-linearity of the ears themselves (thanks to the Fletcher-Munson curves).

You'll then try to compensate for these. If possible, try to cut frequencies first. This is preferable, because all equalizers add phase shifts to the sound when you boost a certain range. This results in unwanted coloration. The more you boost frequency ranges, the more phase shifts usually creep in, and the more difficult it becomes to fit the instrument into the mix. Many sound engineers use their equalizers very sparingly. Nevertheless, the maxim still applies: anything goes! If it sounds good, then it's right.

Golden EQ Rules
- If it sounds opaque and undefined, cut the range by 250 Hz.
- If it sounds nasal, cut the range by 500 Hz.
- Cut it if you want it to sound better.
- Boost it if you want to achieve a different sound.
- You can't boost frequencies that aren't present in the original recording.

Instrument & Magic Frequencies
- Bass drum: Bass at 80-100 Hz, hollow sound at 400 Hz, attack at 3-5 kHz
- Snare drum: Fat sound at 120-240 Hz, attack at 900 Hz, crisp at 10 kHz
- Toms: Fullness at 240-500 Hz, attack at 5-7 kHz
- Floor tom: Fullness at 80 Hz, attack at 5 kHz
- Hi-hat and cymbal: Brass at 200 kHz, sparkle at 8-10 kHz

General Tips
- Use a narrow bandwidth (Q) when cutting and choose a wider Q factor when boosting.
- If you want something to stand out, reduce the low end. If you want it to blend in, reduce the high end.

 

When you make a superlative-like statement and say some plugin is "the best de-esser ever made"; I'm asking about your statement and not the plugin. It has presets in it literally called Smooth De-esser and Fast De-esser; as well as one called Plosive Removal, which makes a lot of sense to have in a plugin also used to deal with sibilance. I'm not in need of a demo, but wondering how you arrive at that statement.

 

Is this some sort of REX2 magnum opus you worketh upon???

 

Just don't forget to use a linear-phase EQ when hi-passing, or you'll get nasty peaks due to sub frequencies changing phase.
Also working with pre-rendered loops is a pain. A proper way to remove rumble and noise is to put gates on individual drums, but since you can't do it in a loop you just have to trust its creator.

 

As a side mention- It's all good and well getting any kind of drum sounds right.
The more important thing is what happens to them - for example - when you put a bass, a couple of guitars, a keyboard, some horns and whatever the the lead is - a vocal or instrument. I say this because while the drums sounds might get to exactly where you want them and it was correctly stated that they can cover the entire frequency spectrum... there will be frequency cancellations in several of the other instruments, or the drums you just spent all that time on when you add the others in.

Mix / Clean up the drums in the full environment you wish to clean them up in. This especially goes for a drum loop. You will likely have to redo a fair bit of what you did if you do them all by their lonesome.

 

Batch processing ?

Until they are the same source, so you can tweak one reference loop and apply it to every other, i don't see any batch processing apart multiband compressor/broadcast treatment for drums.

Because it will need a "generic" treatment, expanding OR compressing each band, with a certain levelling after.

EQ is NOT enough. And can"t be generic/batch.
May be dynamic EQs ... but they are even harder to set.

Some suggestions :
https://www.arturia.com/fr/products/software-effects/mix-drums/overview
https://www.voxengo.com/product/drumformer/

Specific tools that may give good results on batches.
Try some presets until you find a good base, and load a single loop from each source and try it on them.

Ableton/Bitwig are good candidates to check different loops on the same track.