Currently, Oscil::phMod() uses a Q15n16 (or SFix<15,16>) as argument for the phase modulation of an oscillator:
where the fractional n16 part represents almost -1 to almost 1, modulating the phase by one whole table length in each direction.
So, in theory, one do not really need the integer part at all, as modulating by 1 brings you back to the original position, hence modulating by 0. The overflow of a 16bits should be able to reproduce this behavior.
The sign might be a problem, I have not though that completely through yet, but this could be a point of optimization.
Currently,
Oscil::phMod()uses aQ15n16(orSFix<15,16>) as argument for the phase modulation of an oscillator:So, in theory, one do not really need the integer part at all, as modulating by 1 brings you back to the original position, hence modulating by 0. The overflow of a 16bits should be able to reproduce this behavior.
The sign might be a problem, I have not though that completely through yet, but this could be a point of optimization.