1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
|
CLASS:: MedianTriggered
summary:: Median of recent values, triggered
categories:: UGens>Filters
related:: Classes/MeanTriggered
DESCRIPTION::
Calculates the median of the most recent length values, but only paying attention to values input while the trigger is greater than zero. One application of this is to calculate a running median of values coming from FFT analysis.
While trig<=0, the last-measured median is held constant.
The length argument is set at initialisation, and cannot be modulated. The length is directly reflected in the amount of real-time memory taken by this UGen, so please think carefully before using large values of length. Values in the low single- or double-figures are expected.
The median is implemented using a simple "selection sort", which is another argument against using large values for length since the performance is not tailored for large collections.
EXAMPLES::
code::
s.boot;
// Simple polling of median values - you could do this without a UGen!
x = {|val=1, t_trig=0| MedianTriggered.kr(val, t_trig, 3).poll(t_trig, "Median of recent 3 values"); }.play;
x.set(\val, 10.rand.postln, \t_trig, 1); // Execute this repeatedly
x.free;
// Same but with 4 values - note the way medians behave with even-sized sets
x = {|val=1, t_trig=0| MedianTriggered.kr(val, t_trig, 4).poll(t_trig, "Median of recent 4 values"); }.play;
x.set(\val, 10.rand.postln, \t_trig, 1); // Execute this repeatedly
x.free;
// Using it as an audio filter - compare the sounds of these:
x = {WhiteNoise.ar(0.1)}.play;
x.free;
x = {MedianTriggered.ar(WhiteNoise.ar(0.1), 1, 3)}.play; // Note that Median.ar is more efficient for this kind of thing
x.free;
x = {MedianTriggered.ar(WhiteNoise.ar(0.1), 1, 11)}.play; // Note that Median.ar is more efficient for this kind of thing
x.free;
::
|
