thegearforum.com
That is really interesting. Does this change significantly if you pitch down more or less? Experientially, I found that there’s a sort of natural sag that happens with my baritone guitar (tuned B to B), and that the poly pitch fast mode does a great job mimicking that “latency” as compared to other pitch shifters that feel less natural.
Check the latency with peaks BEYOND the first. I expect that’s where you’ll see the difference. The first offset is probably just the necessary shift required to sample frequencies down to the lowest ones so the pitch algo can act on them.
either
An impulse response might not give you a reliable indicator of what you're trying to measure. The pitch algo takes some time to work out what it's doing, and the latency between the note being played vs. the pitch returned might vary as a function of the pitch, too (e.g. lower pitches may have a bit more latency than higher ones). Since an impulse response is an incredibly short pulse, it might not be giving the pitch algo enough to time to do what it does in a meaningful way for measuring latency.
Use a realistic test: record a DI of some actual playing, consisting of varied notes, chords, durations, attacks. Feed the DI into the pitch block with different settings and record the output in your DAW. Then compare latency offsets by identifying key features in the original (e.g. waveform peaks) and see where they fall in the recorded outputs.
