[Stk] Blit error?

[Sean Luke]

Hi Julius!  Thanks for responding.   Maybe, perhaps complex harmonics.  

Playing with some code under the spectrum analyzer, it appears that +1 in fact doesn't go over Nyquist, so it seems to produce the right number of harmonics.  Perhaps the math is right but the description is incorrect instead?

Sean

> On Sep 22, 2020, at 11:06 PM, Julius Smith <julius.smith at gmail.com> wrote:
> 
> Hi Sean,
> 
> 0. Yes, it looks like that should be -1 not +1.
> 
> 1. Maybe the thought was "complex harmonics"?  So, each sinusoid contributes 2 except at dc and fs/2 where it's only 1
> 
> - Julius
> 
> On Tue, Sep 22, 2020 at 6:07 AM Sean Luke <sean at cs.gmu.edu> wrote:
> Hi all from Washington, DC.  I am comparing the code in Blit.cpp against the original paper here:
> 
>         https://ccrma.stanford.edu/~stilti/papers/blit.pdf
> 
> ... and have come across two oddities.
> 
> 0. The paper says that M as: "M is the largest odd integer not exceeding the period P in samples", and describes M as
> 
>         M = 2 * Floor[P/2] + 1
> 
> (and this is what Blit.cpp does) But these two things are not the same.  If P is an even number, then M = P + 1.  That is, it exceeds P.  I think I am misunderstanding something, and was hoping you might either be able to set me straight.
> 
> 1. The paper describes M as "the number of harmonics" (page 5, top left paragraph).  This doesn't seem to be true: if M is close to P, then it must be describing approximately *twice* the "number of harmonics" before we exceed Nyquist.  Indeed Blit.cpp has a variable called maxHarmonics and M is about twice that.  Wanted to make sure the paper's description is incorrect.
> 
> Sean Luke
> 
> 
> 
> _______________________________________________
> Stk mailing list
> Stk at ccrma.stanford.edu
> https://cm-mail.stanford.edu/mailman/listinfo/stk
> 
> 
> -- 
> "Anybody who knows all about nothing knows everything" -- Leonard Susskind