understanding NoiseGate

Noisegate often does precisely what I expect in that it reduces or eliminates heavy breaths, etc. between vocal phrases, but doesn’t eliminate all background noise.

Every so often, it seems to behave differently and in this case it unexpectedly reduces the room noise pauses between words to total silence, which of course sounds weird.

The raw recordings seem similar and I haven’t been able to figure out why I mostly get the first effect but sometimes, the second. Any thoughts?


Noise Gates may be used to cut the level of residual noise between sections of a
recording. While this is essentially a very simple effect, this Noise Gate has a
number of features and settings that allow it to be both effective and
unobtrusive and well suited to most types of audio.

SELECT FUNCTION: Select whether you want to ‘apply the Noise Gate effect’ or
‘test the noise level’, or read this help file (Help 1 - 3 and ‘Tips’).

LINK STEREO will process both channels of a stereo track the same and will only
gate audio when both channels fall below the gate threshold.

LOW CUT FILTER: Removes sub-sonic frequencies including DC off-set. DC corrected
tracking is used whether or not the Low Cut Filter has been selected.

GATE FREQUENCIES ABOVE: applies the gate only to frequencies above the set level

  • may be useful for reducing tape hiss, but may also introduce a small amount of
    ‘phase distortion’. Setting this below 0.1 kHz will switch this feature off.

LEVEL REDUCTION: How much the gated sections are reduced in volume. Gating to
absolute silence may sound unnatural and ‘clinical’. The default -12dB will
provide significant reduction in the audio level of gated sections while
avoiding the ‘gate slam’ that is a common criticism of noise gates. If the Level
Reduction is set below -96 dB the gate will ‘shut’ to absolute silence.

GATE THRESHOLD: When the audio level drops below this threshold the gate will
‘close’ and the output level will be reduced. When the audio level rises above
this threshold the gate will ‘open’ and the output will return to the same level
as the input.

ATTACK/DECAY: How quickly the gate opens and closes. At the minimum (10 ms) the
gate will fully open and close almost instantly as the audio level crosses the
threshold. This could cause the gate to ‘flutter’ or ‘snap’. At the maximum
(1000 ms), the gate will begin to slowly open (fade-in) 1 second before the
sound level exceeds the Threshold, and will gradually close (fade-out) after the
sound level drops below the Threshold over a period of 1 second. Longer gate
times (up to 10 seconds) may be achieved using text input rather than the


The Gate level should be just above the level of the noise to be removed. If the
gate threshold is too high, sound that should be retained will be cut. If the
threshold is set too low, noise will be above the threshold and allowed through
the gate. -6dB (maximum) will cut even moderately loud sounds. -96dB (minimum)
will allow virtually all sound to pass.

For a suggestion of a suitable Threshold setting, select a portion of the track
that contains only background noise, then use the “Analyse Noise Level”

The default Attack/Decay of 250 milliseconds will usually produce good results.
If the noise cut-off sounds too abrupt, increase the Attack/Decay setting.
Longer settings are often better when a high degree of noise reduction is being

If the noise remaining after applying the effect is too great, either set the
Level Reduction lower or raise the Threshold.

Suggested work-flow:

  1. Analyse the noise level
  2. Test the gate on a short selection with the Level Reduction set to -100 dB
  3. Adjust the Gate Threshold if necessary and retest.
  4. Adjust the Level Reduction slider and test. For the most ‘transparent’
    results, avoid setting the Level Reduction any lower than absolutely necessary.
  5. When you are happy with the settings, apply to the entire song.

Thank you, Steve.

It may be obvious to you knowing Noise Gate so well, but I am unable so far to figure out, using a consistent level reduction of -12 dB in vocals that seem to have comparable amounts of background noise, I sometimes accidentally get the quiet parts reduced to absolute silence, while most often the room noise is retained with the heavy breaths removed?

What can I say? I don’t know what settings you are using other than “-12 dB” and I don’t know what audio you are applying the effect to. Perhaps if you post a short audio sample and say exactly what settings you are using then I may be able to offer some suggestion. See here for how to post and audio sample: https://forum.audacityteam.org/t/how-to-post-an-audio-sample/29851/1

Noisegate often does precisely what I expect in that it reduces or eliminates heavy breaths, etc. between vocal phrases, but doesn’t eliminate all background noise.

Sometimes, someone will jump into the middle of a problem with little or no background. Then the forum elves go charging off to solve one specific problem and wrestle with that over pages and miss something obvious by not knowing The Job.

What’s The Job?

I’m recording an AudioBook and I consistently miss the ACX noise reading because of traffic noises in the street,” for one example.

What are you doing?


I appreciate it. I have to wrestle with this more myself, I realize I can’t answer the question intelligently.

I think that some of the voice recordings I work on are coming down almost to zero after noise cleaning between words and so noisegate sounds very natural and works terrifically to remove swallows, etc.

On others the steady hiss of background noise remains too high for noisegate to sound natural, is my feeling.

I have to struggle with more recordings to be surer what my problem is. :smiley:

thanks again!

I realize I can’t answer the question intelligently.

He said, swooping abruptly to the left, making me even more convinced we’re not solving the actual problem.

Nobody ever accused NoiseGate of producing natural, graceful recordings from noisy or damaged presentations. I suspect we’re all going to come out of this with red spots on our foreheads from beating them against a wall—and no show.

If you’re trying to get rid of wet mouth noises, a package has been produced to help with that. If you actually have The Yeti Curse (frying mosquitoes) from a misbehaving microphone, an all but unpublished filter was developed to help with that. If you have straight microphone hiss (ffffffffff), then a gentle Noise Reduction might be in order. All these may produce a spoken-word presentation with much more natural sound than trying to struggle with NoiseGate.

Select some work and post a sample of a raw recording, or record a purpose-built sample.



I find NoiseGate to be personally very helpful when it comes to my own excessive noises like deep breaths or background noises from my desk, but as soon as my dog barks it will pick it up unfortunately. It’s hard to do settings around that since the sound is short and loud :frowning:

Otherwise it works great for me.

sound is short and loud

And furry with a cold, wet noise.

Rex is now part of your show, particularly if he went off during dialog. There’s nothing quite like a quiet room.

So you’re doing something that sartre245 isn’t, or other way to. We should get him to post some of the work and see what the differences are.

You can totally have magic problems. For example, many less expensive microphones and microphone systems don’t filter low pitch noises (thunder, large lorries, earthquakes and some data errors). In most instances, the rumble doesn’t make any difference at all, so why put the expensive filters in? Nobody is willing to pay for them.

The problem is with rumble that comes and goes. You may not be able to hear anything at all, but the Audacity tools and filters may go nuts—when they feel like it. They can hear everything.