Noise Gate for removing microphone spill - keychain idea...

Hi guys

I’m a live sound engineer just starting out using Audacity and editing WAV’s from live recordings.

In situations where the recording was … ahem … less than optomal, and “embracing the spill” from other instruments is just a bit excessive…

… I’m trying to tidy up the toms from the drum kit. All the kit (and everything else on stage unfortunately) has been picked up the drum kit overheads … too hard to do anything with…

… however, the Rack Toms and Floor Tom mics especially are perfectly suited to attempting to remove the “rubbish” they picked up. It’s mostly the electric bass and the lower frequencies of the vocals creating a wash of mud picked up by the microphone, but using a HPF to get rid of them would also rob all the tone out of the drums. Although though the toms are used only a few times during each song, the “mud” the microphone also captured remains for the whole song.

I tried out the Noise Gate plugin with some extreme settings and it did a surprisingly good job in leaving only the tom sounds behind. HOWEVER…
… the 10ms attack time is way too long and misses the initial drum attack
… the resultant 10ms release time (why are they linked? Sigh) clips off almost all of the ring from the drum.

For the attack, I know - because of sampling and processing - 10ms is probably the best I can get with this plugin.

Is it possible to somehow (probably with something else, not this plugin)
… duplicate the floor tom track
… offset the duplicate track by 10ms so it’s drum hits are 10ms before the original
… use the “early” duplicate track to then trigger the noise gate over the original, overcoming the 10ms minimim attack time?

I admit this would only solve the problem with the attack time, the far-too-short release time of 10ms still remains (perhaps if the offset idea works, use 100ms and set the Noise Gate combined attack/release time to that?)


I forgot to add, each mic was recorded to it’s own channel and has it’s own WAV. The whole kit came to 8 mono WAV plus a stereo for o’heads

I’m guessing that you are referring to this Noise Gate: Missing features - Audacity Support
(For optional plug-ins, it helps if you give a link so that we know what you’re talking about.)

I’ll assume that is the plug-in that you are referring to, and answer accordingly.

To make it simpler. The majority of Audacity users have little experience of audio engineering, and for general purpose use it’s enough to just set the “speed” of the gate.

As this is a Nyquist plug-in, it is very easy for more ‘advanced’ users to modify the effect. Nyquist plug-ins are plain text files, so they can be modified with a plain text editor such as NotePad++ (

To make separate attack / release controls, open the Noisegate.ny file in NotePad++ and find the line:

;control attack "Attack/Decay" real "milliseconds" 250 10 1000

Change that line to:

;control attack "Attack" real "milliseconds" 250 10 1000
;control decay "Release" real "milliseconds" 250 10 1000

Then at around line number 143, change from

(setq decay attack) ; this could be replaced with a slider if required


(setq decay (/ decay 1000.0))

That shouldn’t happen because the gate uses “lookahead”. In other words, if the attack time is 10 ms, then the gate begins to open 10 ms before the signal exceeds the threshold, so that it is fully open when the threshold is crossed. Going less than 10 ms is likely to create an audible “click”.
If the attack is being missed, then there’s some other reason.

It’s mostly the electric bass and the lower frequencies of the vocals creating a wash of mud picked up by the microphone, but using a HPF to get rid of them would also rob all the tone out of the drums.

Typically, everything except the kick-drum and bass guitar are high-passed (at maybe ~180Hz). You don’t need deep-bass from the toms and overall you should end-up with a cleaner mix that sounds better on most playback systems when all of those low frequencies are not “competing”. (It’s also common to EQ the kick & bass differently to give each “their own space”, and/or to [u]duck[/u] the bass with the kick.)

You can probably high-pass the overheads at an even higher frequency.

Sometimes the bass guitar is high-passed since most of the bass guitar sound is harmonics. But, I think that would be at a lower crossover frequency… maybe 50 or 60Hz. (The lowest note on a standard bass is ~41Hz.)

And… Don’t judge the sound of the toms (or any other track) by themselves. They may sound wimpy & thin high-passed when soloed but your mix will probably sound better if you allow the kick & bass to “own” the low frequency range.

Although though the toms are used only a few times during each song, the “mud” the microphone also captured remains for the whole song.

If the noisegate doesn’t do the trick, you can use [u]automation[/u] (that would be the Envelope Tool in Audacity) to bring-up the toms only when they are needed.

The thing is… With a noisegate or automation it may begin to sound unnatural when the spill goes up-and-down and you may have to compromise.


Next time… Maybe you can experiment with mic positioning for less spill. You are using directional mics, I assume?

You can find lots of good information by searching the [u]Recording Magazine[/u] for “drums”.

Drums are difficult to record live, but ironically some of the best-sounding drums I’ve heard are on a couple of DVD concerts that I have. It makes me wonder if most studio-recorded drums are “over-processed.” Or, maybe these particular DVDs are not loudness-war compressed…

One of the “advantages” of a live recording is that you can’t go-back and re-record it… You aren’t spending days in the studio re-recording the vocals 'till they’re perfect, etc. It also limit’s the amount of editing you can do (because of the spill) and overall it should go more quickly than a studio recording and you’ll actually finish the project instead of trying to perfect it forever…

Make sure to back-up and permanently archive those irreplaceable WAV files! You don’t want to loos them now, and sometime in the future you (or someone else) may want to do a “re-mix”

Thanks for your input, DVDdoug

It’s not the bottom end/subs that are the problem, it’s the low midrange of the (male-baritone) vocals and bass giving me that “Mud”. Every microphone on stage has it to some degree, I’m trying to do what I can to keep it under some sort of control. The Drum overheads have picked up essentially a full band mix, and it’s pointless to try to do anything with them. That means they already contain the toms so by severly processing the individual tom channels/wav files, I’m just adding some “oomph” to whats already there, hopefully without adding any more mud.

If I was doing the live mix (and it was a bigger room) I would have used gates on the toms anyway - decent hardware gates (LA Audio, BSS, and some DBX) can get the gates to open with values of microseconds. But this recording had none of that - it was each channel into a Focusrite A/D and straight into ProTools with minimal processing on the way in. And this recording will not be repeated. Not with that lineup, not with that material.

I don’t think the envelope idea will work. It’s two sets of 45min of music - that’s a lot of envelope points to set for 90 min of audio, especially trying to squeeze them in between each tom drum hit.

I’m still working on the “NY” “Noise Gate” plugin as having the best chance so far to solve this, and followiing Steve’s idea of editing the plugin settings to get differnt attack and release times. However, the plugin is throwing errors when being called on - it seems to remember now-nonexistant “NY” files and old settings, and disabling/re-enabling doesn’t seem to help. I haven’t yet found a way to refresh the plugin list and reload them.

The only thing I have working for me is that maximin level of the background mud is -15db from 0, but the tom hits are at -6db from 0 or even louder (they cop a fair whack). That’s a handy bit of difference in level that can hopefull be exploited to drive that background mud to even lower. An expander with a 6db boost set to trigger at a -15db threashold (and then use the “Amplifier” plugin to wind the result back 3db) would work. Gates almost do the same thing.



I was able to make the adustments you suggested to the NY Noise Gate - and it works a treat.

The final settings were:

-30db reduction
-12.5db threashold (even higher than the snare hits)
0.2ms attack
1000ms release

The tom hits are still peaking at -6db from 0 but the background “Mud” is now -56db, almost inauadable.

I might try a longer release time, and more level reduction, but at the moment it’s doing what I was looking for, and no audable click as the gate opens.

Thanks for your help, Steve, appreciated.