tID_mean-help.pd

#N canvas 271 134 860 476 10;
#X floatatom 494 283 10 0 0 0 - - -;
#X msg 494 192 927.55 819.1 527.85 497.31 303.98;
#X text 37 120 Suppose you take 32 spectral centroid measurements over
the course of a 2 second sound \, and 16 centroid measurements over
the course of a 1 second sound. A distance between these two feature
vectors cannot be calculated because they have a different number of
components.;
#X text 37 290 Creating a larger feature vector composed of several
time-varying timbre features summarized in this fashion will pack a
great deal of information into one compact descriptor. In other words
\, you can create a vector that describes how spectral centroid \,
flatness \, flux \, and individual Bark-frequency cepstral coefficients
vary over the course of a sound.;
#X text 531 350 Also see:;
#X text 37 200 Taking the mean and standard deviation for each of these
centroid lists creates two 2-component lists that CAN be compared to
compute a distance. All the precise detail of how centroid varied over
the course of each sound is lost \, but mean and standard deviation
do provide meaningful general information about a data set.;
#X text 37 30 [tID_mean] calculates the arithmetic mean (average) of
a list of numbers. The list must have at least 1 element. It is intended
to be used for summarizing time-varying timbre attributes that are
measured with other timbreID externals. This provides one way to compare
the spectro-temporal characteristics of sounds with different durations.
;
#X obj 1 -7 cnv 10 400 10 empty empty empty 20 12 0 14 -233017 -66577
0;
#X obj 1 -7 cnv 10 10 400 empty empty empty 20 12 0 14 -233017 -66577
0;
#X obj 494 238 tID_mean;
#X obj 603 350 tID_std;
#X connect 1 0 9 0;
#X connect 9 0 0 0;