posted 07 April 2014 15:03            

partialPanning_v1.kym

based on the gustl´s very inspiring spectral smoothing thread here at the forum i experimented with partial panning using the wire between.
my goal is to place/move individual partials between the left and right channel. for now iam trying to do it by random without interpolation between successive values.

(to avoid any confusion i will use the term channel as a synonym for output or speaker while the term leg refers to the amp or freq part of a frame).

i made a patch using 2 oscbanks (one for the right and one for left channel) which is not working correctly at the moment.
the basic idea is create random amplitudes (left leg) for all partials in a frame of one channel by using scaled noise output (0-1), while simply take the inverse
of the noise output for the other channel.

the modulation signal (for the product with the left leg) needs to create 252 random samples each 252 samples with the option to hold these random samples for a longer period. iam doing this by feeding white noise into a memory writer (252 samp duration and set to silent mode) plus
a sample player that reads from the wavetable (frequency: 252 samp).

of course this is a very basic and maybe complicated way of doing it in kyma and iam very curious about your ideas and critics.
also maybe someone has an idea why the panning sounds blurry, with no sharp left or right panning and doesn't seems to change pannings when writing and reading new noise samples into/from the wavetable…

is this because the 2 oscbanks are not in sync?

posted 07 April 2014 17:26            

partialPanning_v2.kym

I've added two variations. I won't explain them but will let you ask questions, if you need to.

BTW the odd sound you are hearing is because SOS with missing partials do sound odd. You can tell that it is working because if you mono it up outside Kyma, it sounds normal.

posted 08 April 2014 05:03            

both of your variations are straight forward. i specially like the second version with the smoothed noise, that let the partials move slowly (with feedback-values close to 1)

i made the panning by random just to hear how it sounds to do it on partial-basis instead of doing it on bin-basis.

from what i know there is no gui in kyma that let you manually control the pan position of muliple (e.g. 252) partials. something like the multislider object in max…
or did i miss it somewhere?
it would be great if you could simply draw into a wavetable and than read from it ...
this would be great for partial filtering as well-using a filtercurve drawn in the wavetable.

posted 08 April 2014 06:39            

probably you are right.

but its definitely another horse to control parameter that effects the curve instead of drawing the curve directly. anyway, both have pros and cons.

posted 10 April 2014 15:14            

partial_panning_v4.kym

regarding my patch:
i used capytalk/ smalltalk to create the arrays for the xvalues and values parameters and change the their values on the fly. here i choose 12 xvalues, with a squared distribution. in my case the xvalues are placed at the following partials: 0, 1, 4, 9, 16, 25, 36, 49, 64, 81 etc.
to these fixed xvalues i add a random amount for variations on the x axis.

my problem:
i realized a strange jittering of the curve in the oscilloscope which also seems to effect the sound in the form of amplitude modulation. the best way to see and hear it is when you bring down the bpm speed to below 30 bpm. sometimes it also freezes the movement of the curve movement totally. but the cpu of my pacarana seems to be fine…
does anyone know why this happens?

posted 12 April 2014 15:02            

i tried it with vmin and vmax but it doesnt change anything.
i guess that it is a capytalk syntax thing that i don´t understand.

i made a simplified comparison between 2 kinds of YVallues array constructions,
while using -1, 0,1 for XValues:

version 1:

{|inter|
inter := !Speed bpm s.
(1 to: 3) collect: [: i | i inverse * ((inter tick) nextRandom) smooth: inter]}

version 2:

{ ((((!Speed bpm s tick) nextRandom) smooth: !Speed bpm s }
{ ((((!Speed bpm s tick) nextRandom) smooth: !Speed bpm s }
{ ((((!Speed bpm s tick) nextRandom) smooth: !Speed bpm s }

any ideas why the first version does that jittering and second doesn´t?

thanks!
ciao, johannes

posted 13 April 2014 13:54            

compare_XenOsc_randomDrawing.kym

i realized that, but forgot to change it... anyway, it doesnt seem to cause the jittering.
but guess what, after replacing nextRandom with nextRandomWithSeed the movement of the curve was nice and smooth. i append a patch that shows it.

so is it a nextRandom related behavior?

posted 13 April 2014 19:01            

seeded

posted 14 April 2014 13:44            

partial_panning_v5.kym

i append two patches doing randomized partial panning.
one is using a squared and the other a fibonacci based mapping.
within both patches i tried to accomplish very slow changing partial pannings.
(try the presets)

i made the experience that partial panning will be best observed via headphones
or in acoustic treaded rooms.

feel free to criticize. iam very curious about your feedback.

thanks for all your help.
ciao, johannes