another look at Brightness Analyser

[elmiquito]

Would it be possible to take another look at the Brightness Analyser module ? I know it uses fiddle~ to measure the spectral centroid as a weighted mean of all the frequencies in the signal, but somehow users find it counterintuitive as perceptually brighter sounds don't reliably correspond to higher frequencies of the brightness parameter. Maybe it's something to do with the analysis window size, or the speed of the metro object controlling the rate at which the brightness values are generated?