BLIT oscllators are a bit late

[Andreya-Autumn]

Or should that be "a blit late"? :P

Anyway:

Classic seems to start late by a varying amount somewhere between 40 and 80 samples, wavetable ditto by about 80-160 samples.

S&H Noise is interesting. With Correlation at max or min it behaves exactly like Classic. But with correlation at default it seems like it's accurate! As accurate as the MIDI block processing will let it that is. Maybe there's a clue there somehow?

[baconpaul]

Lemme see if I can replicate this in a regtests as a first step

[blancoberg]

I stumbled upon this bug as well. To me it looks like instead of starting on a random phase, it instead delays the signal for 0-1 cycle. So the delay varies depending on what note you are playing. On lower notes the delay can be quite substantial

[mkruselj]

That's a plausible explanation!

[baconpaul]

Yes that’s exactly what they do and what the test above shows.

the fix to in init start them at start phase - 1 then run them up to zero but when I did that they clicked like crazy at low attacks.