[Fpga-synth] Signals and Sample Rates

[Dennis Bemmann]

Am 17.05.2008 um 18:43 schrieb Scott Gravenhorst:
> Eric, I like your use of the term naive - it describes some of what  
> I did - my sawtooth NCO
> wave is just the upper bits of the phase accumulator.

The things you have to be careful about also depend a lot on the kind  
of signal you are processing. If it's just signals from digital  
oscillators, then it's much easier than with samples. For example if  
you have a naive sawtooth or square oscillator, you already know that  
it doesn't generate critical frequencies except for one place: the  
discontinuity. So you can treat the discontinuity and you're fine. I  
got really good results even with no oversampling.

It's a lot more tricky if you're processing samples, because your  
algorithm doesn't have that kind of knowledge about the signal and  
frequencies above nyquist can occur with every single output sample. I  
don't know how aliasing filters for samples work. With wavetables it's  
a bit easier because you can pre-generate several aliasing-free waves  
for different pitches.

Cheers,
Dennis