Hey hey,

working with the fantastic PVS opcodes again, I experimentally worked out the delay to perfectly synchronise the original audio input with the PVS output. Using pvsanal and pvsynth respectively, with an FFT and window size of 1024 and an overlap of 256, I arrived at )1024 + 512) samples delay.

Now why the 512? My intuition was to have the extra 256. Are the other 256 independent of the analysis parameters or linked to either of the three parameters defining the FFT analysis?

I think I asked about this before, but somehow didn't fully appreciate the answer. Though that was probably some years ago.

Best wishes,

Jeanette