[sdiy] Observations of synthesized stretched-harmonic waveforms and subjective comments on their musical qualities

[cheater cheater]

> How difficult would it be to construct the additive synthesis scheme that
> you used in your simulation with analogue blocks? Or is that on the same
> craziness level as a 40-band analogue vocoder?

I think not difficult at all. The problem with a 40-band vocoder is
that the bands need to precisely hold frequency otherwise they will
make the sound overshoot and distort, as well as you need a level
meter for each band, and so on.

The formula for the frequency is very simple as you see. You could use
a single expo / current source, and then use that with as many cores
as you want. The important thing is they don't even have to be that
exact, because you'll be mostly using stretched-harmonic spectra
anyways - so no problem if the frequencies are off by a bit. Maybe you
would need some current mirrors for each core, who knows - I'm not
that good with electronics.

Instead of having a resistor from the current source to the cap you
would have a resistor and a VCR FET or something like that. The static
resistor would change from partial to partial and be fixed to make the
core work at n*f_B if the VCR is at some 'neutral' setting; all VCRs
in the bank would get the same CV. When that CV raises, the VCRs would
respond with the same way but hopefully - and that's not really
thought through but very possible - they would shift the frequency of
the partials so that they stop being a harmonic series. I think this
could give you a 'polynomial' style stretch or something similar,
maybe an 'exponential' style stretch if the FET is operating in the
non-linearity.

If before the junction that goes to the VCR and real resistor you add
another VCR for each core, it could probably change the 'linear
stretch' l.

Then after the summing node of that VCR and real resistor for
'polynomial' style stretch you could tap in another current source,
which is a single one shared across all cores, and which sets the
linear offset f_o.

Then a simple sine shaper after the triangle core takes care of the
rest, although it would be interesting to be able to hear a
synthesized sound have its partials morph from sine to triangle.

Finally a mixer with an op amp for each partial and some clever way of
doing the cv for the 'w' parameter and some autogain makes the
waveform more mellow or more brittle.

Cheers
D.

On Thu, Apr 8, 2010 at 16:44, Ian Smith <taciturn_unquiet at hotmail.com> wrote:
>
>> Yep, that's why I hope to one day be able to do this in analog - it
>> would definitely sound more interesting than digital.
>>
>> D.
>
> How difficult would it be to construct the additive synthesis scheme that
> you used in your simulation with analogue blocks? Or is that on the same
> craziness level as a 40-band analogue vocoder?
>
> -Ian (the more knobs the merrier) Smith
>
> ________________________________
> The New Busy is not the too busy. Combine all your e-mail accounts with
> Hotmail. Get busy.