Request a mathematical/logical basis for a situation come across

Dear illustrious Leaders and senior members,
Suppose I have an audio file with segments of background noise, the maximum RMS of the noise segment being -64dB.
By experience I observed that if copied the track to make another, and I use the customised popmute Nyquist command and set the Threshold only at -50dB or above (in dB, moving from more -ive towards lesser -ive, 0 and +ive), only then the noise profile and RMS value would be identical as the original track. Otherwise, not.
What is the logical and mathematical basis for this difference in the value around 14dB for effectively filtering out the background noise with pop and clicks?

PopMute is the wrong tool for trying to clean -50dB noise.
PopMute is for attenuating occasional loud (> -10dB) pops, like on a scratched vinyl record.

pop mute attnuating vinyl scratch

Dear Mr. Trebor,
I thank you for posting your reply with the accompanying gif file for me. I appreciate your support.
Assuming that the attached gif file is for a stereo track, please have a look at my modified PopMute values.
Then my method.
From one mono track, I copy to create another mono track. Apply the modified popmute over the second track. Invert the second track. Then Mix the two tracks to make another mono track without those noises. I then truncate the mono track and finally get only meaningful information left.
Masterstroke: I eliminate the noise with itself.
Briefly described in my immediately earlier Topic.
And I really admire my insight because I stand on the shoulders of giants like you, and a great tool called Audacity.
I continue to get outstanding results. Clean meaningful audio, without noise.
The only problem I now face is that I don’t have a mathematically pre-determined value for the threshold value every time for every unique audio and the set of a valid reason behind such choice of values.
The reason and the mathematical reason behind it.

There’s not going to be any noise-reduction where PopMute has muted: no cancellation.
You may as well use a noise-gate to attenuate/silence the noise where there is no signal only noise.
https://manual.audacityteam.org/man/noise_gate.html

[PopMute is an upside-down noise-gate].

I have to again reiterate that it works because I invert the second track after the modified popmute is applied on it. Then I use Mix and Render applied on the two tracks to create a new track.
Please try yourself with the modified popmute values, as mentioned earlier.
For the last few days I have been successfully using this procedure, measuring each track and the results with Analyse/Measure RMS.

So it seems that I have made an innovation/invention, as I usually do in my core field.

Please. Follow the steps yourself. With the Modified PopMute values that I used, not the usual popmute.py.

In the end, please don’t forget to appreciate me for this innovation, Mr. Trabor.

I really admire you all for your support and allowing me to stand upon your shoulders to learn, experiment, learn more and look further all the time.

My method works the best for downloaded clips of audio podcasts. One could for instance test with youtube postcasts such as:
youtu.be/MGQUQs29aco,
youtu.be/fCxnvim_S2o,
youtu.be/uhnUwowZhf0,
etc.
I have been able to reduce time, noise, improve quality, etc., of the clips I required extremely well with my method and truncating silence, speeding up the tempo and applying distortion to level up the audio.

A belated (counting from 4th Dec 2023) Welcome to the Forum.
The thread is Complete in terms of the information desired.
Best.

This topic was automatically closed after 30 days. New replies are no longer allowed.