What's the latency on the vocals in a VoiceLive 3?
There's two types of latency in our vocal processing systems, "Dry" and "Effected".
The DRY vocal path, if Lead Delay is OFF, is a constant 6ms. This means that regardless of the active vocal effects your lead voice always has 6ms of latency, which is below the threshold of perception (disregarding distance from a speaker/monitor which adds 1ms per foot as sound travels through the air).
The WET, or effected vocal path has variable latency depending on which effects are active. Harmony is the most costly algorithm and adds roughly 24ms of latency.
What this means is that your lead voice will go through the system in 6ms and the Harmony voices will come out 18ms (24ms-6ms) later. While this is perceptible if you were listening to copies of identical sounds, it's much less noticeable when it's "other voices" generated by the harmony block.
If you would like to have your DRY and WET voices lined up perfectly (perhaps if you're recording and can nudge the track after the recording pass) then you can turn Lead Delay to Auto, which will compensate for any effect latency and delay the dry voice to match.