[pd] FFT sine to saw oscilator (was:[pd] hanning window + fft)

[Charles Henry]

Actually, you inspired a different idea from me...Not a solution...but
something I thought would sound weird.
without exponential decay...the impulse response of a simple recursive
filter that repeats itself every 3 samples, for example, would look
like
1,0,0,1,0,0,...
then, 0,1,0,0,1,0,... on the next block
then, 0,0,1,0,0,1,... on the 3rd block

If these were in successive overlapping windowed blocks, it would be a
sound that rotates through all the frequencies (in the fourier basis)
over 3 blocks...
but it requires that we have a feedback from every 3 samples, and to
experiment, it would be good to be able to vary the number of samples
in feedback

Now if you want to create a sawtooth, you have to compute the fourier
series to see the coefficients.  and phase is important.  the sawtooth
function is odd, so it's entirely a sine series.  To get the phase
correct, you have to feed all your frequencies to the imaginary input
of ifft~

Chuck




On 9/7/06, Chuckk Hubbard <badmuthahubbard at gmail.com> wrote:
> On 9/7/06, Charles Henry <czhenry at gmail.com> wrote:
> > On 9/5/06, Chuckk Hubbard <badmuthahubbard at gmail.com> wrote:
> > > Just a thought... the output of a recursive filter from an input of an
> > > impulse would be several impulses of linearly changing amplitude at
> > > equally spaced intervals, and it would seem like that ought to give
> > > some sort of sawtooth if you put it through IFFT.
> >
> > I tried to do this, like you said, but couldn't figure it out.
> > I tried to use z~ with a throw~ and catch~ on very small blocksizes,
> > but I couldn't get the throw/catch to work.  same with send~/recieve~
> >
> > then, I ran the output to ifft~ on overlapping blocks and windowed
> > with the hanning window
> >
> > Let me know if you get it working
> >
> > Chuck
>
> I was thinking something like this.  I was wrong, it's exponential
> decay, not linear, so it wouldn't create a sawtooth exactly.  I also
> can't for the life of me figure out why the impulse response moves
> every time I hit bang on here.  It seems like it should always start
> right at the left side of the graph, but sometimes it doesn't show up,
> and other times it skips around in the graph.
> But if you set coefficients for the filter, bang until something shows
> in the response, and set the volume, it does make a harmonic sound.
>
> -Chuckk
>
>
>