@manuels OK Thankyou for suggestion.

Thanks for explaining.
The idea of using the filter to find the 'right' bandwidth is really interesting. I’ll take a look at the plot you shared. Do you think this method works well with data that has different densities?

I find the idea of using a zero-phase filter or a moving RMS window really interesting for histogram smoothing. I'm curious how would a weighted average approach compare to something like kernel density estimation for smoothing?

I think the problem is that I’m not using enough different numbers. I’ll try using longer lists and change the settings to get more random results. Let me know if I need to check anything else.

That’s an interesting point.
The bubbling sound might happen because both FFT and MDCT work with overlapping windows and process data in a similar way. This could be why it sounds like low-bitrate MP3s