Random frequency modulation (was [sdiy] Interesting f vs. t graphs of pitch instability)

[Tom Wiltshire]

On 7 Oct 2008, at 00:04, Scott Gravenhorst wrote:

> Tom Wiltshire <tom at electricdruid.net> wrote:
>>
>> Having a knob on an oscillator that goes from cool digital sterility
>> at one end, via CEM3340 to drifty moog, and then finishing at Andre's
>> wavering MS20 sounds great to me. The question is what signals do you
>> need to feed into the oscillator and how to generate them  
>> efficiently.
>> Your graphs represent some actual data, which is very helpful indeed.
>>
>
> Interesting you mention this, My GateMan FPGA synths all have noise  
> modulated pitch,
> it is something I really like.  Mine go from subtle and slow drift  
> up through tonal
> noise to almost just plain noise.

Yes, this is a feature that my SH101 has that I think the Pro-One is  
missing. One day I'll get round to feeding the noise source into the  
mod panel - it's a simple enough mod.

I'm interested to know about your noise modulation, Scott. Do you  
feed the noise into a logarithmic (v/oct) or linear modulation input,  
and what depth do you use? Presumably this is calculated per-sample,  
and at some pretty high rate too. I've tried the same trick at lower  
sample rates (50KHzish) and it doesn't sound as good. The sound  
becomes "spattery" rather than smoothly drifty.
Consequently, I've also been experimenting with other types of drift  
waveform, including random linear slopes (pick a random point, slope  
towards it, when we arrive pick another, and slope off there too) and  
cosine interpolation between random points. These are quite nice in  
that they have an underlying "frequency" (how often the points get  
picked) which gives you an element of control. They seem to mimic  
slow drifting quite well, though (like some old analogues) it can be  
hard to make things sound in tune!
The advantage for my situation is that such slowly changing waveforms  
still sound ok at lower sample rates.

I'm quite sure there are lots of other varieties of random signal  
that one can feed into a digital oscillator's frequency, phase, or  
amplitude mod inputs to warm things up. Any ideas appreciated.

Thanks,
Tom