[sdiy] A crazy idea..

[Fredrik Carlqvist]

Of course, the cool thing with the vocoder-type solution is that you can
use any music or sound to generate control voltages for your modular.

Rough component count for composite-controlvoltage encoder and decoder:

2 4051
2 generic binary counters
2 555-type one-shots
1 555-based oscillator
4 NPN transistors
3 diodes
9 buffer op-amps
Some resistors and capacitors

I realize that 10-12 bits of resolution might not be enough for VCO
control, but it should do for the other control voltages. Maybe the
pitch CVs come from MIDI anyway? :-)


Fredrik C


-----Original Message-----
From: owner-synth-diy at dropmix.xs4all.nl
[mailto:owner-synth-diy at dropmix.xs4all.nl] On Behalf Of harrybissell
Sent: den 6 oktober 2005 03:46
To: Jeff Farr
Cc: Eric Honour; synth-diy at dropmix.xs4all.nl
Subject: Re: [sdiy] A crazy idea..

You could use multichannel FM as well... but that would get kind of
tricky.
Switched capacitor filters would be a good choice here. But watch out,
as higher order filters carry their own problems like group delay etc.

Might be easier to do CV-midi conversion and record the data in
something like Cakewalk (or whatever you kids are using these days :^).
I used to stripe one channel with SMPTE time code and lock synths to
that. Po' man's multitrack.

H^) harry

Jeff Farr wrote:

> I was thinking about 8 bands per audio channel, starting around 16k on

> downward in octaves.  However, I realize this will take a steep BP 
> curve to decode so that there is enough headroom between bands for 
> good resolution, I'm not sure what kinds of slope can be easily 
> achieved but I imagine having 12 or 24db of  space for each octave may

> not be high enough for a good resolution 'decode'.
> On 10/5/05, Harry Bissell Jr <harrybissell at prodigy.net> wrote:
>
>      > On 10/4/05, Eric Honour <autophage at gmail.com > wrote:
>      > >
>      > > What about using a channel as one half of a
>      > vocoder?
>      > >
>      > > IE, one audio channel carries twenty different
>      > CV's by having twenty
>      > > different (harmonically unrelated) tones of
>      > varying amplitude... much like
>      > > the 'ghosting' technique previously mentioned, but
>      > not actually containing
>      > > usable audio (it'd just sound like a crappy
>      > dissonant chord) - then using a
>      > > series of very narrow bandpass filters into a
>      > bunch of envelope followers?
>
>      You don't need 'harmonically unrelated' tones if you
>      are using sine waves (which have no harmonics). The
>      limiting factor would be how steep you can make the
>      bandpass filters... and how stable the recording
>      method is (wow, flutter, speed changes ?)
>
>      The lower audio tones will STILL be much slower to
>      recover.  I'd think that twenty tones would be really
>      pushing the limits of what you could do with practical
>      bandpass filters...
>
>      H^) harry
>