Filtering Out Audio Below a set dB?

Hello community. Thank you for always helping with any question I have before I ever need to ask it :smile:

Today is different. I know this has been asked in some ways before, but I wanted to know if there was a way to utilize the current audio filters to completely wipe out all sound below a certain -dB threshold. Some describe this as a ‘noise filter’ but if there is a way to do it with the existing filters I would love to know. I routinely use the existing noise gate to limit low and high frequencies but this has a markedly different purpose (filtering frequency vs volume).

I usually capture audio through recording software that has this filter built in (it’s “Action!” and it is adjustable from -40dB to 0, I think), and can’t hear the results of a given setting until after the video is finished recording. Having the ability to adjust on-the-fly in Shotcut would be fantastic. :smile:

(ps. some have suggested Audacity in the past. My workflow is time-sensitive as it is and that software introduces too many steps for me to sacrifice time for quality!)

Shotcut have a Noise Gate audio filter.
It’s a very flexible filter because you can filter by frequency, the range and from which you want the audio to be heard, the detection threshold, the time in milliseconds for the filter’s reaction (on and off).

Additionally, you can use the low pass filter and high pass filter, for example, to remove high frequencies from high-frequency noise.
You can also be guided by the audio spectrum oscilloscope.


In Audacity, you can perform a noise removal, after the audio is captured, by analyzing an area where there is only background noise.


Thanks for the reply :slight_smile: this unfortunately does not solve my question – I hope my question was clear enough in that I already know about the noise gate filter and the use of audacity.

I’m looking for filtering out audio below a certain threshold of volume, specifically, if it exists.

I routinely use the existing noise gate to limit low and high frequencies but this has a markedly different purpose (filtering frequency vs volume).

(ps. some have suggested Audacity in the past. My workflow is time-sensitive as it is and that software introduces too many steps for me to sacrifice time for quality!)

I definitely don’t understand the question. Are you asking about an audio noise removal filter?
Something perhaps based on an analysis of a zone of the audio?
A subtraction filter as in a DAW?
Are you looking to remove noise that may have gotten into the recording process?
I’ll take a closer look.

Nothing so complex. Just to keep it simple, I just need a setting or filter that would allow me to say, in the most general sense, “if the volume dips below -40dB [or whatever number I set], don’t play the sound at all.”

I was wondering if the existing compression filter could be used for this purpose, or if another filter had a set of parameters that would allow me to use it in this way.

This is exactly what Noise Gate does by adjusting Threshold. The default frequency range is very wide.

So to audio software is frequency (hz) the same as volume (dB)?

I ask out of genuine curiosity, I specifically use Mirilis Action! to record and the noise cancellation filter it uses seems to give completely different results than the filters I’ve used in shotcut. The one in Action seems to work by literally cutting out all sound below the threshold if it is quieter versus it being a particular frequency, so when I place my microphone in different areas relative to my mouth I can hear swallowing when it is close versus no swallowing when it is farther away yet I haven’t adjusted the parameter. At the same time, this seems to have no effect on the quality of the audio that I can discern.

On the other hand, using the high or low pass filters in Shotcut, as well as the noise gate, seem to have an effect on the overall quality of the audio resulting a loss of some bass tone to my voice which I’d rather like to keep. That’s why I assumed they worked differently.

That said, sometimes I have to use that for its intended purpose, but it can’t be used in substitute for the filter Action comes with. That’s why I wanted to know if Shotcut inadvertently had something similar. :slight_smile:

I think you are confusing a noise gate filter with a noise cancellation filter. They are two different things.
The noise gate filter does not let the signal pass when it falls below an adjustable dB value and allows the signal when the dB level rises from that value + hysteresis or threshold. It is also possible to adjust the reaction time of the filter since it detects the set value and its effect on the output audio. The response curves also affect the result of blocking unwanted audio signals (steeper = fewer milliseconds, softer = more milliseconds)
This is useful so that certain noises (a few dogs barking away, for example) are not recorded.

Noise cancellation does not work like that.
It is based on the phase change of two audio signals in which certain non-coincident frequencies are canceled.
In this case, there is no limitation on dB but subtractive processing on a signal and its anti-signal.
The frequency range in which this noise cancellation will be focused is normally oriented to cancel unwanted noise during a voice recording.
There are a few audio filters in Shotcut that might be able to do something similar by making a combination.
I mean Copy Channel, Mix Channel, Balance, and others. But this would not be like directly applying a phase inversion to one of the channels and would require more steps (I suppose many more steps than using the inversion in Audacity for example)

In the Steve Harris LADSPA filter collection, there is a plug-in for this, but I don’t know if it is implemented in Shotcut.

No confusion here, I understood from the beginning they were two different things. :slight_smile:
I knew my ignorance of the subject would otherwise prohibit my asking clearly, though.

So, I apparently work with a noise gate filter in Action, and just need to know if something similar can be achieved in ShotCut, by using any combination of filters (the noise gate filter as it is titled in Shotcut seems to work on a different principal). I don’t mind it being a bit of a process to set up, as I save a template file with all of channels and filters set up and work from that as a basis with each video I edit. But again, just for clarity, it is the noise gate that I need, as you explained it – something that works with dB.

I need to keep from recording the dog barking, so to speak. Audacity is as usual just not an option.

If the answer is “no” or even “not sure” I can accept that. I might be able to find more information from the MLT documentation, and perhaps make my own filter. That would be an educational challenge.

This has been answered before by Dan (Shotcut).
The Noise Gate filter in Shotcut works based on dB (yes) but also incorporates the possibility of adjusting the frequency range (Hz) where we want the filter to detect the volume level (again dB)
I don’t know what filter, what functionality, or what the filter you use in Action is based on. It seems to work well for you and that’s what matters.
What does it mean that the noise gate filter in Action works differently from the noise gate filter in Shotcut?
What does it mean specifically?

If you want to avoid the audio of a distant dog barking, or someone talking quietly in the next room, or even the sound of a distant bell, using the Noise Gate Audio Filter (so-called in Shotcut) can be helpful.
The trick is to adjust the values accordingly.
If the noise is very loud, it will force you to adjust the dB value to high values as well, so a decrease in your voice volume level will be cut off and will not work well for you.
Note that as long as there is an audio signal above the set dB value (your voice and noises or barks), the filter will do nothing, because it only cuts the signal when the dB level is below the setting.
An example: you are talking and at the same time there is a dog barking, the filter will let your voice and the dog’s sound pass through, but since the bark has a lower volume, your voice can effectively mask the barks. But the barking is still there.
The moment you stop talking, the volume of the audio decreases (to the level of the dog’s bark). The dog continues to bark, but if you adjust the dB threshold properly, the filter will cut the audio signal until the audio volume does not increase again.
Anyway, I am explaining the operation of a noise gate filter, but in the links, there is already a technical explanation about this.

It may be that the inclusion of a range of frequencies may confuse how to use the filter.
Actually, if you set the low-frequency key to the minimum and the high-frequency key to the maximum, you will be telling the filter, to take into account, for the dB measurement, the maximum frequency spectrum (33.6 Hz to 23520 Hz).
The information provided by Dan regarding the functionality of the Noise Gate filter in Shotcut does not differ in any basic way from Audacity’s Noise Gate filter whose functionality and workflow you can find here:

And while there are more effective and elaborate (and more costly in time and money) ways to restore audio or remove certain unwanted sounds, processing with Audacity is by no means a bad idea. It just depends on the expected result, your skill with the software, and the time you can devote to this process.

All of the Steve Harris plugins are available in Shotcut builds, but we have NOT added a UI for all of them.

So what’s the name of the audio phase inversion filter in Shotcut? I can’t find it.

I edited the sentence above to include the important word “not”. You can contribute it.

1 Like

Yes, I checked the Ladspa folder in Shotcut and saw the whole collection of filters.
I looked at the link to contribute, but I don’t know anything about programming, sorry.

When I read this I knew instantly what you mean. :slight_smile:
This is what “Compressor/Limiter”-Filters in order with a noisegate-filter do.
But in SC there’s no ‘opening-gain’ in the Noise Filter.

Can you explain this a little more?
I’d like to understand what you mean and what effect it has on the audio.
I know about the compressor but not about the opening gain associated with the noise filter.
It is always good to learn something new. :slightly_smiling_face:

Are you familiar with OBS-Studio?
The Noise-Gate has that ‘gain-in threshold’ that opens at a specific amout of -xxdB.

1 Like

I only use OBS to record the screen while playing Rocksmith 2014. I never add filters to avoid latency.

Thanks for the information. I will look at OBS what that filter does.

Yesss! I knew I wasn’t going crazy. :joy:
Can you advise how I might best emulate how OBS does it using multiple filters in SC?
If I can get it to perform close enough, it might help me begin recording without the filter of Action, so that I can rely fully on SC and edit sound in post.