ok, that makes sense. I did already create something using delete, as I'm applying a dynamic color to the arrays after they've been created. But having a look at the the partial-trace makes more sense.
Best,
Joao
I think I see what happened - the recorded trace reports 0 dB but this is the output of a 'rmstodb' conversion in the patch - I presume sigmund~ is finding a peak at amplitude less than 1/10000 which then gets converted to zero.
Something similar was happening with your earlier test runnign phasor~ into sigmund~ - there are very quite peaks in phasor~ output because of foldover, and sometimes sigmund~ picks them up, sometimes not.
If you want to suppress very quite traces from the partialtracer patch, you can alter the add-trace abstraction to consider a trace empty if it's below an amplitude threshold and then to start a new trace if a trace then goes from below threshold to above threshold. Alternatively, now that it's possible to delete scalars using th pointer object, you can make something that culls very quiet traces after the fact.
cheers Miller
On Sun, Apr 12, 2020 at 11:56:38AM +0200, Jo??o Pais wrote:
Here is a txt with the tracing I did using the doc patch - a contrabass tone. The 0-amplitude tracks appear when there is more silence, in this case after the 2nd half of the sample. I produced this file by sending a "write" to pd-trace-list, with load you can load it into the same patch.
Best, Joao
Am So., 12. Apr. 2020 um 04:07 Uhr schrieb Miller Puckette msp@ucsd.edu:
It's possible to have nothing in a track - in that case, the flag (the 4th item in the list sigmund~ outputs) should be -1.
If that flag is saying there's a sinusoid of amplitude zero, something's wrong with sigmund~.
cheers Miller
On Fri, Apr 10, 2020 at 11:29:35PM +0200, Jo??o Pais wrote:
Hello list,
I'm trying out [sigmund~], and wanted to confirm something related to results I'm getting. When using the patch doc/4.data.structures/14.partialtracer (and also a derivate patch of
mine),
I get lots of tracks with 0 amplitude as result of the analysis. The settings are [sigmund~ -t -npts 1024 -npeak 40 -maxfreq 8000 peaks tracks] in the help file, and [sigmund~ -t -npts 2048 -npeak 200 tracks]
in
my patch.
When comparing the result of the analysis with a spectrogram of the
sample,
the frequency content does match - the tracks with 0 amplitude don't
exist
in the sample, i.e. they're silence.
My doubt is, is it normal to have tracks with 0 amplitude? Or is there something wrong with the analysis settings? (if these tracks are a derivative of [sigmund~], that's fine, I can
filter
and delete them afterwards)
Best,
jmmmp