Hi, sorry to hear the latency is a showstopper for you.
Cleanly separating transient and tonal material inherently requires introducing latency. Why? To identify something as tonal, you need to know it is stable and repeating. How do you do that? By looking at the signal over time, which means there is going to be a delay before you can pass audio.
We’ve had some thoughts about how to reduce this latency, but don’t have a timeline on when that might happen.