Hi, Oleg.

I'm not sure if we're looking at different things but the IDE outputs two
identical channels with the test code that you posted below, unlike the two
sets of samples that you posted.

Yes, better mostly as it is a higher-order one, but I remember Niemitalo
using a genetic algorithm to get the coefficients. I'm not sure if that's
equivalent to de Soras (or whatever they based their implementation on) or
which one is better at the same order.

Best,
Dario

On Wed, Sep 2, 2026 at 10:31 PM Oleg Nesterov <[email protected]> wrote:

> On 09/02, Oleg Nesterov wrote:
> >
> > Just one note for now, I can hardly understand your image.png, because
> you
> > plot both magnitude and phase in the same window. But this doesn't really
> > matter, I'll send another email tomorrow.
>
> Yes...
>
> but before I sleep, let me write the really stupid but simple test case,
> see it below.
>
>         $ rdsp /tmp/TEST.dsp -rn 44100 | tail -10
>         +0.704864134613         +1.00002277294
>         +0.451158937770         +1.00002181207
>         +0.241717523054         +1.00001909207
>         +0.0934312453347        +1.00001483230
>         +0.0182593254887        +1.00000937631
>         +0.0222643446729        +1.00000316413
>         +0.105123299911         +0.999996696755
>         +0.260153654137         +0.999990495791
>         +0.474852279770         +0.999985061337
>         +0.731903830271         +0.999980831679
>
> so it seems that the code you showed in your email was indeed wrong.
>
> But you know, this doesn't really matter...
>
> I only can't agree with "here is a better one". I mean, I hope that now it
> is
> clear that both versions "do the same". Just Niemitalo's version uses
> order=4,
> while your version uses order = 16/2 = 8.
>
> And hiir library allows to design the "same" filter with any
> order/transition.
>
> Oleg.
>
>
> -------------------------------------------------------------------------------
> import("stdfaust.lib");
>
> // copy-and-pasted from your email
> combAP(coeff, x) = f ~ _ with {
> f(s) = x'' - coeff * x + coeff * s';
> };
> coeffsHilbertVicanek = waveform {
> 0.0406273391966415,
> 0.2984386654059753,
> 0.5938455547890998,
> 0.7953345677003365,
> 0.9040699927853059,
> 0.9568366727621767,
> 0.9815966237057977,
> 0.9938718801312583,
> 0.1500685240941415,
> 0.4538477444783975,
> 0.7081016258869689,
> 0.8589957406397113,
> 0.9353623391637175,
> 0.9715130669899118,
> 0.9886689766148302,
> 0.9980623781456869
> };
> coeffHilbertVicanek(i) = coeffsHilbertVicanek , int(i) : rdtable;
> hilbertVicanek(x) = (x : seq(i, 8, combAP(coeffHilbertVicanek(i)))) , (x :
> seq(i, 8, combAP(coeffHilbertVicanek(i + 8))) : mem);
>
> // I _think_ this fixes your code. At least makes it simpler and more
> effecient ;)
> fixed = H(a1), H(a2)' with {
>         a1 = 0.0406273391966415, 0.2984386654059753, 0.5938455547890998,
> 0.7953345677003365,
>              0.9040699927853059, 0.9568366727621767, 0.9815966237057977,
> 0.9938718801312583;
>         a2 = 0.1500685240941415, 0.4538477444783975, 0.7081016258869689,
> 0.8589957406397113,
>              0.9353623391637175, 0.9715130669899118, 0.9886689766148302,
> 0.9980623781456869;
>         H_sect(coeff, x) = f ~ _ with { f(s) = x'' - coeff * x + coeff *
> s'; };
>         H(as) = seq(i, outputs(as), H_sect(ba.take(i+1, as)));
> };
>
> mag2 = _,_ : ^(2) + ^(2);
>
> process = os.oscrs(1000)
>         <: hilbertVicanek, fixed
>          : mag2, mag2;
>
>
_______________________________________________
Faudiostream-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/faudiostream-users

Reply via email to