dan b

I've been thinking more about the question in this post

https://cycling74.com/forums/neural-oscillator-sample-wise-iterative-calculations-troubleshoot)

and decided the question is a little different and warrants a new topic.

I'm trying to use iterative algorithms to produce audio - equivalent to producing audio from the Lorenz equation or something (though with more internal variables)

I did this in PD using expr~ which calculates on a vector basis - the time base of the vector being the system buffer length. I'm now trying to implement in MAX in windows, which has no expr~ object. So I'm using the maths objects. It works in PD, it doesn't work in MAX. Since I'm using the math objects in MAX I'm wondering if the time base of the calculations might be the difference.

So - my question is; does anyone know how MAX audio math objects calculate - do they work on a vector-by-vector basis or on a sample-by-sample basis?

Also, in either case can anyone tell me if a chain of audio math objects is evaluated from beginning to end before MAX moves to the next sample/vector, and whether the presence in the chain of patchers effects this calculation ordering? Those kind of things will have a big effect if I'm trying to do sample by sample/ vector by vector calculations.

I've been thinking more about the question in this post
(https://cycling74.com/forums/neural-oscillator-sample-wise-iterative-calculations-troubleshoot)
and decided the question is a little different and warrants a new topic.

I'm trying to use iterative algorithms to produce audio - equivalent to producing audio from the Lorenz equation or something (though with more internal variables)
I did this in PD using expr~ which calculates on a vector basis - the time base of the vector being the system buffer length. I'm now trying to implement in MAX in windows, which has no expr~ object. So I'm using the maths objects. It works in PD, it doesn't work in MAX. Since I'm using the math objects in MAX I'm wondering if the time base of the calculations might be the difference.

Also, in either case can anyone tell me if a chain of audio math objects is evaluated from beginning to end before MAX moves to the next sample/vector, and whether the presence in the chain of patchers effects this calculation ordering? Those kind of things will have a big effect if I'm trying to do sample by sample/ vector by vector calculations.


question-about-audio-math-objects-and-ordering-in-audio-calculation-chains

You could set the vector size in a [poly~] subpatch, but this gets CPU intensive pretty quickly.

i am not using it myself, because i require mac OS9 compatibilty, but there

is a 3rd party expr~ object which should do everything the pd object does.

i am not using it myself, because i require mac OS9 compatibilty, but  there
is a 3rd party expr~ object which should do everything the pd object does.


Vanille- I think you understood exactly- you certainly answered the question. I wouldn't be surprised if you're right and C ends up being the answer, the paper I sourced the idea from implemented in C.

Alternatively, I guess I could accept the lower resolution and calculate vector wise rather than sample wise - I'd just have to set my delay to the buffer size. I'll give that a try too.

Luke - that's a good idea, I'll give that a crack.

Roman. Thanks, I was under the impression that they haven't been compiled for windows though. I'll check the dl when I get onto a PC with max on it.

(it's here if anyone else needs to find them 

http://www.yadegari.org/index.php?option=com_content&task=view&id=89&Itemid=88)

Thanks for your help all -
Vanille- I think you understood exactly- you certainly answered the question. I wouldn't be surprised if you're right and C ends up being the answer, the paper I sourced the idea from implemented in C.

Roman. Thanks, I was under the impression that they haven't been compiled for windows though. I'll check the dl when I get onto a PC with max on it.
(it's here if anyone else needs to find them http://www.yadegari.org/index.php?option=com_content&task=view&id=89&Itemid=88)


Question about Audio Math objects and ordering in audio calculation chains