| I'll take a WAG at this one. I notice the same effect in Cool Edit Pro 1.1
when I select adjacent areas (to the sample!) of a sound and filter them
differently. Apparently the filter algorithm (any of them, so far as I can
see from the math behind them) doesn't guarantee to leave the endpoints
fixed, or zero, or any known value. So you get discontinuities at the
boundaries of the area the filter's being applied to, which, obviously, you
hear as clicks. Anyway, while that may not be exactly what's causing your
particular problem, it's a problem inherent in digital filters. I've only
ever coded FFT-based filters, but I'm somewhat familiar with the theory
behind IIR's and FIR's. Maybe someone else has a different take on this.
If it's what I think it is, you might try to modify your instrument to do a
forced zero or fast crossfade on every note. Off the top of my head, I
don't know if there's a way to make CSound's filters do that for you, but it
wouldn't be hard to tack on top.
On a totally different note, can anyone tell me how to desubscribe from the
list? I've lost the instructions I had, and it doesn't appear to behave
like other listproc-style lists I've seen. Maybe I just don't remember the
correct server name. Anyway, I'm relocating, and until I have a new ISP,
I'll have to forgo the list.
Oh, yeah. Goodbye to the guys up at the U of U Music Department, those of
you who still remember me .
> -----Original Message-----
> From: owner-csound-outgoing@maths.ex.ac.uk
> [mailto:owner-csound-outgoing@maths.ex.ac.uk]On Behalf Of Jonathan
> Drexler
> Sent: Tuesday, February 02, 1999 10:33 PM
> To: csound@maths.ex.ac.uk
> Subject: Filter clicks revisited
>
>
> I posted something about this earlier but am not sure it got through
> to the list properly. I have been having a problem with filter
> clicks where the filter cutoff-freq is made to track the pitch of
> the note. This has been a problem particularly in legato
> passages. Here is some sample code which illustrates what I
> mean: it plays 2 legato scales. The first is passed through a
> filter (notice the clicks at the beginning of most of the notes).
> The second bypasses the filter and does not click.
> I wondered whether anyone out there has run into this
> problem and could help me figure out a solution.
>
> Thanks to all.
>
> *********
> ORCH
> ********
> sr = 44100
> kr = 44100
> ksmps = 1
> nchnls = 2
>
> instr 10
> ;p4=decibels
> ;p5=octave.pitch
> ;p6=wavetable
> ;p7=phase
> ;p8=filter stack
> ;p9=switch: bypass filter if set to 1
>
> iAmp = ampdb(p4)
> iNote = cpspch(p5)
> a1 oscili iAmp,iNote,p6,p7
>
> if p9=1 goto BypassFilter
> aFlt1 tonex a1, iNote*2, p8,p7
> outs aFlt1,aFlt1
> goto Endlbl
> BypassFilter:
>
> outs a1,a1
> goto Endlbl
>
> Endlbl:
> endin
>
> **********
> SCORE:
> *********
> f10 0 8192 10 1000 00 040
>
> ; Dcb note tbl phase FltrStack BypassFilter
> i10 0.00 0.50 80 8.00 10 0 4 0
> i10 + . . 8.02 . -1 . .
> i10 + . . 8.04 . -1 . .
> i10 + . . 8.05 . -1 . .
> i10 + . . 8.07 . -1 . .
> i10 + . . 8.09 . -1 . .
> i10 + . . 8.11 . -1 . .
> i10 + . . 9.00 . -1 . .
>
> i10 6.00 0.50 80 8.00 10 0 4 1
> i10 + . . 8.02 . -1 . .
> i10 + . . 8.04 . -1 . .
> i10 + . . 8.05 . -1 . .
> i10 + . . 8.07 . -1 . .
> i10 + . . 8.09 . -1 . .
> i10 + . . 8.11 . -1 . .
> i10 + . . 9.00 . -1 . .
> e
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> !
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> !
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
> !
>
>
> |