hi i'm new here and i have a couple of questions

i'm workin on an audio/video performance. i'm using pd to send parameters to the video guy with netsend.
i want to analyze my audio and send him the env~ values of 4 frequency bands (low, mid-lo, mid-hi, high)
but i don't know how to do this. i tried with lop~, 2x bp~ and hip~ on the signal but one-pole filters are not good to clearly divide the spectrum in 4 bands.

what can i use? are there some externals or patches that could help?

i thought to netsend~ too, leaving the analysis part to the video guy, but i don't know if the video sofware's (vvvv) netreceive is able to handle it.

thanks!

More filters. Stack em up in series to get a sharper rolloff.

obiwannabe, do you mean something like

[inlet~]
|
[bp~]
|
[bp~]
|
[...]
|
[outlet~]

all with the same bandwidth and center frequency?

Exactly. Each filter you add increases the steepness between the passband and the stopband. Because resonance is not required (and undesirable in some analytic applications) it might be good to stack lowpass and highpass filters.

[hip~ 100]
|
[hip~ 100]
|
[lop~ 300]
|
[lop~ 300]

to make a 200Hz bandpass filter with a width of 200Hz

@obiwannabe said:

[hip~ 100]
|
[hip~ 100]
|
[lop~ 300]
|
[lop~ 300]

to make a 200Hz bandpass filter with a width of 200Hz

Can someone explain why we have to use two hips and lops for each frequency band?

As far as i know, the hip and lop are 1-pole filters. So the more you stack, the more steeper the cutoff goes. But then i might be totally wrong, and somebody else will give a correct answer.

I made something simmilar for a frequencyanalyser some time ago using ordinary bp~. After reading here i changed them into lop~ and hip~ At least in the lower frequencies it help with getting rid of some noise, but still im not realy pleased. They all output to similar values in the end.

Anyone got sugestions? Should i use more filters, would it help to narrow the bands?
Is this even the best way to make a frequencyanalyser?

Edit: I experimented some more using an osc~ now and at 8 lop~&hip~ pairs i get quite decent bleeding. Somehow i get output in the higest band at certain input frequncies, gues it has something to do with Nyquist.

Edit2: Also compared it with 8 bp~ but scince smaller bandwith also gets mor oscilation from them it doen't give good results.

Conclusion: Use lop~&hip~ pairs and a lot of them (I guess you could set a lower bandwith and use fewer).

http://www.pdpatchrepo.info/hurleur/spectrum_analyzer.pd