posted 09 May 2014 10:26            

iam looking for a way to do the modulo operation at samplerate. to accomplish that i need samplerate division which is not part of kyma at moment.
is there any alternative way to a sr based modulo or something similar than modulo like wrap?
it would be great to have a sr modulo and division module in kyma.

posted 09 May 2014 11:32            

ClippingDivide.kym

It would be great if you made some examples to help explain to people how it works.

posted 15 May 2014 13:41            

phaserand@cartesians.kym

i tried to implement and it works but now my dry fft-ifft signal sounds already a bit phase randomized. i guess this is because of the complex multiplication. even if my random-phase-amount is zero, a slight modulation is happening due to the calculations on the fft, right?
is there something i can do to obtain a unmodulated sound when the phaserand amount is set to zero?

the whole complex calculation i made looks very complicated to me. maybe there is a more easier/ efficient way in kyma using less modules.

THANKS, jo

posted 15 May 2014 15:34            

It looks like there are some problems with your complex multiplies:

* the Difference part has its inputs in the wrong order

* the second Product input to the Sum should have IMAG and Cosine as its inputs (you were using IMAG and Sine)

posted 16 May 2014 17:53            

SRWRAPMODULES.kym

Attached are some sample rate wrap around modules.

It uses the Equality module to do the logic and a scale and offset to do the arithmetic manipulation, but beware, as there seams to be a bug in the Equality module where by if the input B is more than 1 below the input A, the output goes to plus one, regardless of the tolerance setting. I'm carefully choosing which input to use to avoid the fault area and if the module gets fixed it should not alter the operation.

Each set of modules is only able to do one wrap around so you may need to string them together to make it warp around many times. You application will determine how many you need.

We have to avoid clipping so I made it wrap at 1/5 of full scale, so if you want full scale type clipping you would have to attenuate your inputs by 1/5 individually before you mix them together and feed them into the wrapper, and then give it a gain of 5 on the output to get the correct level back. You can of course choose a different reduction factor by altering all the number 5s in the modules but be very aware of clipping at every point of the signal path.

The Wrapping mixer with gain in Petes DSP modules took care of any clipping anomalies and didn't need multiple stages but we don't have that luxury here I'm afraid.

Also make sure you remove the test input constant and the test output S to G controller or it won't be at sample rate.

BTW you can change it to modulo (where it wraps round to zero instead of matching minus value) by changing the 2/5 to 1/5 in the scale and offset modules and you will need an off set of 1/5 but only in the minus wrappers and change the minus tolerance to value 1, but I'm sure you can work this out.

Hope it helps and hope it makes sense