Lorenzo T

Hi all, as a proof of concept and a pedagogical exercise, i wanted to recreate in gen~ a few classic (and not too complex) filter designs. 

I started with biquad succefully but i am having a problem with the SVF chamberlin filter.

Am i perhaps missing something from the msp patching feedback rules in gen? is the function wrong?

SVF_gen_tests.maxpat

Hi all, as a proof of concept and a pedagogical exercise, i wanted to recreate in gen~ a few classic (and not too complex) filter designs. 
I started with biquad succefully but i am having a problem with the SVF chamberlin filter.

Am i perhaps missing something from the msp patching feedback rules in gen? is the function wrong? 

Hi all, as a proof of concept and a pedagogical exercise, i wanted to recreate in gen~ a few classic (and not too complex) filter designs. 


svf-in-gen-works-as-a-codebox-but-not-in-msp-patching-style

Yes, you can't make foodback loops in msp~ that are shorter than the dsp buffer size in your audio settings, so generally, it's not possible to make filters in msp, and it was this feedback limitation on msp that was the original reason pretty much that cycling74 made gen~.

the rule is basically the same, you just have shorter vectors in gen~ (1 sample, where in poly~ 2 is the minimum)

Im sorry to have to correct someone, but Roman is wrong. There is always single-cycle feedback in gen~, including in poly~ objects, via the History operator. Maybe he means there is some change in the functionality of delay() or other built-in methods, which I wouldn't know about, but the results of one audio cycle are always available in the next.

that´s what i am saying, it is always 1 sample.

in MSP you can not go below a vectorsize of 2 samples, which means that the CPU usage is outrageous and you have to introduce one vector of delay 

so as a result the patching looks quite different, but the background is not. you always have to wait until t-1 is available. :)


that´s what i am saying, it is always 1 sample.

i think he was talking about the difference between MSP and gen.

in MSP you can not go below a vectorsize of 2 samples, which means that the CPU usage is outrageous and you have to introduce one vector of delay yourself while in gen~ you´d use history. 
so as a result the patching looks quite different, but the background is not. you always have to wait until t-1 is available. :)




@lorenzo: you are sending "freq $1" to [param f1]


wow, i neede a while.

@lorenzo: you are sending "freq $1" to [param f1]


Thank you very much for your time and for checking the patch!

this was the problem in the msp "graphic" style gen~ patching version, the order of substractions.

And yes.. sorry for the wrong param freq name, it wasnt the problem during the testing stage anyway because i just changed it to match the codebox version "too" quickly before sharing the patch here, i apologize!

It was a useful exercice for me to understand that it could be translated in an msp graphic style without other kind of allowed feedbacks constraints to follow. I ll keep practicing !

(i am attaching the -corrected maxpatch here)

SVF_gen_tests-corrected.maxpat

Thank you very much for your time and for checking the patch! 

H = L - in1 - q1*d1 
with 
H = in1 - L -  q1*d1 

this was the problem in the msp "graphic" style gen~ patching version, the order of substractions.

And yes.. sorry for the wrong param freq name, it wasnt the problem during the testing stage anyway because i just changed it to match the codebox version "too" quickly before sharing the patch here, i apologize! 

i believe the last [+] at out 4 can be removed now. (while patching it is great to always include them)

SVF in gen works as a codebox but not in MSP patching style