Jukka's dashboard

Correct. The 'window' is calculated for every point.

For every frequency data point, 'hi' and 'lo' index points are calculated at 1/2n octave distance on either side of the point. Then, the average is calculated within that range.

Thank you for your answer!

However, this raises one more question. How can you be sure that the window spans always a relative bandwidth (in log scale / 1/n oct)? Is this window purely recalculated for every point in f-domain?

Thanks for your time! :)

If you're in an anechoic chamber, why would you bother using smoothing to begin with!? :)

The algorithm applies a simple averaging routine in the frequency domain data. There are no filter banks involved, which the reverberation time plugin requires.

Each point is calculated as an average of the values surrounding it on either side. The width of the averaging window is kept constant (within 1/n octave) throughout the entire curve.