Improved Noise Gate

[fader8]

This is a simple technique that improves the effectiveness of noise gating a track such as a kick drum when you want to eliminate snare or cymbal leakage.

 

The attached LP8 project file contains a short snippet of kick track that has a lot of leakage. For comparison purposes, I've included an example on track 3 gating the usual way. If you listen to track 3 alone, you'll hear some cymbal leakage while the kick hits. Bypass the Gate and you'll hear it all. Now mute track 3 and listen to tracks 1 & 2 together. Note that the Gate on track 1 has the exact same settings that the Gate on track 3 had, except for the reduction parameter.

 

You'll notice that the cymbals are significantly reduced.

 

Here's the technique:

 

Duplicate your kick track. Send both to the same bus. Leave the fader at 0 for both. Name one Direct and the other Reversed.

 

On the Reversed track, instanciate a Gate plug and set it's reduction to +20 (so it's an expander). Then instanciate a Gain plug and click Phase Invert.

 

Tweak the Gate as you would normally, Leaving the reduction at +20. I've found 0dB hysteresis and about 6mS of lookahead to be ideal but this may change depending on content.

 

Once you get it really good, leave these tracks alone and put any additional processing for the kick on the bus. If you do add plugs to these tracks, make sure they have identical settings.

 

Of course, this isn't limited to kick drums.

 

Sidenote: The Logic Pro 8 Noise Gate is mislabeled in two ways. First, it is a gate/expander, not just a gate. Second, the Side Chain filter section really should just be marked Key Input because it does read from the track audio if you haven't selected a sidechain.

Kick Gate.zip

[Roger Paolo]

And here is my tip to do this

 

the file is named kick gate RP

 

duplicate your drum track, call the 2nd one trigger or something. Route this to a bus.

 

Put a gate on the first track side chaining to the same bus. Make sure that the aux that is created has no output.

 

Extremely cut the side chain high cut to where the kick has the most energy (in the song attached down to 200 Hz), or it will also open on snare drums and hats etc. I have to say that tom sounds can be a pain in the ass when using this technique, but most of the time there is only a few so I usually cut them out manually.

 

fiddle around with treshold, attach, hold and release until you like the sound.

 

Reading your footnote brings me to an idea which I am going to try.. maybe Ill be back with a new tip.

Kick Gate.zip

[fader8]

Hi Roger,

 

OK, you're example illustrates my point regarding the mislabeling of the Side Chain functions. You don't have to create a bus. You can accomplish the same thing on a track alone because the Side Chain filters work with either key input, the sidechain if assigned, or the track itself if you don't assign a sidechain. This is how most dynamics plugs work anyway.

 

So your example is exactly the same as my track 3 "Usual Way" which also sets the hi cut to 200Hz, except you've had to raise the threshold another whole 6dB which takes away from the kicks sound.

 

Compare that to the reversed polarity dup track method and you'll hear the difference. More of the kicks attack and decay are preserved. while reducing the cymbal leakage even further.

[Roger Paolo]

yep, that was the thing I thought of after reading your footnote.... the only thing I can think of that would make sense is to eq the trigger signal in a way that the frequencies to which the gate responds are emphasized, so that the gate responds better to it.

 

And youre absolutely right, in the example you made you have reduced spilled noise more and left more of the original sound intact. I could not hear a lot of difference on my iPod earphones in the train, but I certainly can on my pods at home....

[Matt Mayfield]

 

Now wait a minute...

 

I can see and hear that this does work, but.... how?

 

I understand phase cancellation, but why wouldn't these two techniques do exactly the same thing?

[fader8]

 

Now wait a minute...

 

I can see and hear that this does work, but.... how?

 

I understand phase cancellation, but why wouldn't these two techniques do exactly the same thing?

 

It is a bit of a brain-tease isn't it. The difference is in how expanders do the math on the signal above the threshold. We don't know "exactly" what math's being done, but you can think about it like this:

 

If you're expanding by a factor of 5, a sample value of 2 will end up as 10. If the sample value is 20, it ends up as 100. So the delta gain of 8 (your cymbal wash) is a lot smaller than the delta gain of 80 (your kick).

 

In other words, the higher the signal started out above the threshold, the more its multiplied. The cymbal wash is just above the threshold so its gain isn't multiplied by nearly as much.

 

Capiche?

[OddTime]

Oh that's clever! Very cool. Thanks for posting that!

[ski]

Wow, that's impressive! Very very clever Fader 8!

 

Question tho...

 

With PDC set to Audio and Software Instrument Tracks, it takes a lookahead of 5.8 ms to get the two tracks to become phase coherent so that the non-kick signal cancels. What I don't understand is why, with PDC set that way, it's necessary to introduce a delay into the chain with look-ahead to get the two signals to be phase coherent...

 

Secondly, here's some strange behavior...

 

1. Tracks 1/2 unmuted

2. Set PDC to off

3. Change look-ahead to zero

4. Playback

 

The sound at this point will be essentially the same as before, with the tiniest bit of snare leaking through. Just a very, very tiny amount. But why's it different at all?

 

5. with look-ahead still set to zero, turn on PDC (audio/software inst trks)

6. Playback. Sound is still the same (kick only, teensy bit of snare, all other sounds cancel). But this can't be right though, because previously it required a look-ahead setting of 5.8 ms to achieve essentially this sound.

7. While the track is still playing, move the look-ahead to 5.8 ms. The sound will take on a comb filtering effect. This is strange again, because now we're back to the original state of the noise gate at step 1, before we started messing with anything.

8. Stop, then playback again. The sound will be normal. Why now and not right when we moved the look-ahead value in Step 7?

 

I smell a bug...

[fader8]

Thanks!

 

With PDC set to Audio and Software Instrument Tracks, it takes a lookahead of 5.8 ms to get the two tracks to become phase coherent so that the non-kick signal cancels. What I don't understand is why, with PDC set that way, it's necessary to introduce a delay into the chain with look-ahead to get the two signals to be phase coherent...

 

OK, back up for just a sec. The lookahead isn't used to bring the signals into alignment. It's only purpose is to provide the plug-in with some advanced notice so that it can do its job better. The more time the plug has to prepare for the operation, the more precise and smooth it will be in performing it. PDC then compensates for the delay caused by this preprocessing.

 

 

Secondly, here's some strange behavior...

1. Tracks 1/2 unmuted

2. Set PDC to off

3. Change look-ahead to zero

4. Playback

The sound at this point will be essentially the same as before, with the tiniest bit of snare leaking through. Just a very, very tiny amount. But why's it different at all?

 

Because with lookahead turned off, the plug can't do nearly as good a job.

 

7. While the track is still playing, move the look-ahead to 5.8 ms. The sound will take on a comb filtering effect. This is strange again, because now we're back to the original state of the noise gate at step 1, before we started messing with anything.

8. Stop, then playback again. The sound will be normal. Why now and not right when we moved the look-ahead value in Step 7?

I smell a bug...

 

Not a bug. All plugs report their processing delay to the PDC at the moment playback starts. Making any changes to a plug, during playback, that affect its processing time won't be reported until you stop and play back again. All PDC systems work this way as far as I know. So when you're increasing the lookahead value, you are delaying the signal, thus the comb filtering, but the PDC remains clueless about it until you stop and play again. Then everything's back to hunky dory.

 

As you know, since this is a mixdown technique, you'd have PDC enabled anyway. In fact, stuff like this just isn't practical without it.

[ski]

...this is a mixdown technique...

 

Mixdown? Hmmm... Isn't it better to leave tracks in a virtual and unfinished state so that you can take endless pleasure in imagining how good it'll sound later?

 

[dkgross]

that's what I usually do...

 

[fader8]

Hmmm... Isn't it better to leave tracks in a virtual and unfinished state so that you can take endless pleasure in imagining how good it'll sound later?

 

Yup, then start again, do another version, then another. That way you never finish and you just continuously bill the client until you get the whole pile of cash.

[sallyxi]

nice info,thanks...

[davidpye]

Good technique well spotted.