[sdiy] [synth-diy] numerically controlled superoscillator without hard sync

[Roman Sowa]

And the crossfade between mips can be determined by tuning voltage that 
drives VCO controlling the sample rate.

Is it possible that for example at 1V control step the sample rate VCO 
goes up only by half octave and sample rate converter goes from 512 to 
384 samples so actual output frequency is one octave higher?

In this particular case 10 octaves is no longer 50kHz to 50MHz of sample 
clock range, but only up to 1.56MHz. Easier for VCO.

At 2 octaves higher we switch from 512 to 256 wavetable as a base and 
that should not be audible as far the shorter one is derived from 
512-sample table using the same sample rate converter.
OTOH sample clock is no longer locked to fundamental frequency, and all 
imaging artefacts may appear as good as at constant cample rate 
synthesis. Although I think this would be only noticeable at the 
audiophilistic level.

Or go one step further and do all interpolation using sample rate 
converter, switching between tables at each octave. So we have here 
typical constant rate wavetable player with multisamples.
I guess I'm no variable rate guy...

Roman


W dniu 2014-02-10 10:59, Tom Wiltshire pisze:
>
>
> You *could* do mip-mapping with a variable rate, and you don't have
> to have a sudden change of timbre. My own experiments with it suggest
> that cross-fading from one mip-mapped table to the next is far
> preferable to switching. But it all uses more instructions.
> Nonetheless, there will be some timbre change, partly from the table
> storing different data (that's the point of mip-ampping, right?) and
> partly from the harmonic distortion falling in a different place. If
> that change is gradual, I guess it becomes a "feature" - we call it
> "mojo" not "sloppy engineering", remember?
>
> T.
>