alexandros

I'm trying to use biquad~ without the filtergraph~ (so I can cotrol it with a midi controller) by inputing the minimum and maximum coefficients I get from the list of coefficients coming out from filtergraph~'s left outlet, in biquad~'s 3rd, 4th, 5th and 6th inlets, through a zmap object (so I can scale the controller's values).

The result I get though, is not at all the same as if I use filtergraph. Am I doing something wrong?

Hi,
I'm trying to use biquad~ without the filtergraph~ (so I can cotrol it with a midi controller) by inputing the minimum and maximum coefficients I get from the list of coefficients coming out from filtergraph~'s left outlet, in biquad~'s 3rd, 4th, 5th and 6th inlets, through a zmap object (so I can scale the controller's values).
The result I get though, is not at all the same as if I use filtergraph. Am I doing something wrong?
Thanks,
Alex


coefficients-for-biquad

for sure, because minimum and maximum is not enough, it is not at all linear.

the picture below shows one way how to calculate the coefficients for a pass

filter (which i think is similar to what filtergraph does.)

usually you will prefer an interpolated/smoothed signal as input for [biquad~],

which is why you should look into [filtercoeff~].

pre maxmsp 4.3 users can use [hr.filtercoeff~] ... or embed a small [filtergraph~]

the picture below shows one way how to calculate the coefficients for a pass
filter (which i think is similar to what filtergraph does.)

usually you will prefer an interpolated/smoothed signal as input for [biquad~],
which is why you should look into [filtercoeff~].

pre maxmsp 4.3 users can use [hr.filtercoeff~] ... or embed a small [filtergraph~]
object in an abstraction.
 

The coefficients for biquad~ are interdependent, and they are calculated by filtergraph~ (or filtercoeff~) based on the filter type, gain, frequency, and Q you provide. So you should use your MIDI controllers to control 

 parameters, and feed those to filtergraph~ or filtercoeff~. To avoid clicks, you'll probably want to interpolate between changes using line~ or number~, as shown in this example.

http://music.arts.uci.edu/dobrian/IAP2010/examples.htm#Ex20

The coefficients for biquad~ are interdependent, and they are calculated by filtergraph~ (or filtercoeff~) based on the filter type, gain, frequency, and Q you provide. So you should use your MIDI controllers to control those parameters, and feed those to filtergraph~ or filtercoeff~. To avoid clicks, you'll probably want to interpolate between changes using line~ or number~, as shown in this example.

http://music.arts.uci.edu/dobrian/IAP2010/examples.htm#Ex20


this is an old post, but my problem is very close to this topic: filtergraph~ (for cascade~) is very CPU demanding, especially when params messages are [line 0.]...

So the solution (perhaps) is to use directly coefficents into cascade~ (or biquad~) without filtergraph~.

http://www.earlevel.com/main/2011/01/02/biquad-formulas/)

I tried to build a little example-patch to verify the formulas, but I must admit that the result looks odd: changing params directly on the filtergraph~ with the muose gives different results than passing directly through the formulas, and the worst thing is that the inspector doesn't recognize that the 3 filters are peaknotch (it says display!)

Please, could you help me to find the mistake?

Hi everyone,
this is an old post, but my problem is very close to this topic: filtergraph~ (for cascade~) is very CPU demanding, especially when params messages are [line 0.]...
So the solution (perhaps) is to use directly coefficents into cascade~ (or biquad~) without filtergraph~.
In a webpage (http://www.earlevel.com/main/2011/01/02/biquad-formulas/) there are the coefficents formulas.
I tried to build a little example-patch to verify the formulas, but I must admit that the result looks odd: changing params directly on the filtergraph~ with the muose gives different results than passing directly through the formulas, and the worst thing is that the inspector doesn't recognize that the 3 filters are peaknotch (it says display!)
Please, could you help me to find the mistake?
Thanks


...perhaps Display means nothing more than filtergraph~ is used as display only?

I set the following values for the 3 peaknotches: GAIN 9 FC 87 Q2, GAIN 9 FC 397 Q2, GAIN 9 FC 2024 Q2.

When I set these values directly dragging the mouse on the filtergraph~ the resulting coefficents message says:

(1.003445 -1.996091 0.992804 -1.996091 0.996248 1.073582 -1.840507 0.846273 -1.840507 0.919855 1.015189 -1.980275 0.968268 -1.980275 0.983456)

Instead, when I set the values through the formulas (using filtergraph~ as display only), the resulting message says:

(1.005617 -1.993668 0.988204 -1.993668 0.993822 1.025341 -1.968974 0.946786 -1.968974 0.972128 1.120701 -1.790142 0.746542 -1.790142 0.867243)

Well... it's not SO different, but there is actually a difference.

Here an example based on my patch:
I set the following values for the 3 peaknotches: GAIN 9 FC 87 Q2, GAIN 9 FC 397 Q2, GAIN 9 FC 2024 Q2.
When I set these values directly dragging the mouse on the filtergraph~ the resulting coefficents message says:
(1.003445 -1.996091 0.992804 -1.996091 0.996248 1.073582 -1.840507 0.846273 -1.840507 0.919855 1.015189 -1.980275 0.968268 -1.980275 0.983456)
Instead, when I set the values through the formulas (using filtergraph~ as display only), the resulting message says:
(1.005617 -1.993668 0.988204 -1.993668 0.993822 1.025341 -1.968974 0.946786 -1.968974 0.972128 1.120701 -1.790142 0.746542 -1.790142 0.867243)
Well... it's not SO different, but there is actually a difference.
What do you think about?


i think that the filtergraph object calculates some shorthand and/or low res, it is an old object.