DPCM is Differential pulse code modulation, and its possible to get an estimate of derivative with the right predictor. While the difference between the last two messages is an approximation of the derivative its is a really bad one so extrapulating on 4-points is much better. (dsp theory is really not my area but I'm sure some people will be able to explain much better about dpcm) What I'm looking for is a way to: (1) store the last 3 values in the series. (2) an estimate of the derivative should be x0 - 1/2 * x1 - 1/4 * x2 -1/4 * x3 (where x0 is the most recent element) (3) switch places so the x0->x1 x1->x2 x2->x3 ready to do it again.
I'll try looking at list-abs (I think I tried poking around those on my office mac but pd-extended couldn't find them so I left it at that).
thanks
Hallo, Oded Ben-Tal hat gesagt: // Oded Ben-Tal wrote:
DPCM is Differential pulse code modulation, and its possible to get an estimate of derivative with the right predictor. While the difference between the last two messages is an approximation of the derivative its is a really bad one so extrapulating on 4-points is much better. (dsp theory is really not my area but I'm sure some people will be able to explain much better about dpcm) What I'm looking for is a way to: (1) store the last 3 values in the series. (2) an estimate of the derivative should be x0 - 1/2 * x1 - 1/4 * x2 -1/4 * x3 (where x0 is the most recent element) (3) switch places so the x0->x1 x1->x2 x2->x3 ready to do it again.
Okay, that's not too hard to do. Attached is a solution using [expr], you could of course also use the other math-objects in Pd. I converted list-lastx.pd from [list]-abs into a subpatch here to avoid that dependency.
Ciao
Frank Barknecht _ ______footils.org_ __goto10.org__
hello,
you can use the FIR filter of the mapping lib sending a [1 -0.5 -0.25 -0.25< message on the right inlet to set the coeficient should be ok.
cyrille
Oded Ben-Tal a écrit :
DPCM is Differential pulse code modulation, and its possible to get an estimate of derivative with the right predictor. While the difference between the last two messages is an approximation of the derivative its is a really bad one so extrapulating on 4-points is much better. (dsp theory is really not my area but I'm sure some people will be able to explain much better about dpcm) What I'm looking for is a way to: (1) store the last 3 values in the series. (2) an estimate of the derivative should be x0 - 1/2 * x1 - 1/4 * x2 -1/4 * x3 (where x0 is the most recent element) (3) switch places so the x0->x1 x1->x2 x2->x3 ready to do it again.
I'll try looking at list-abs (I think I tried poking around those on my office mac but pd-extended couldn't find them so I left it at that).
thanks
PD-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
-
cyrille henry -
Frank Barknecht -
Oded Ben-Tal