Hey @vlh, thanks for the suggestion!
You might want to take a look at this related thread as well, where a very similar idea is being discussed.
At the moment there’s no built-in way to handle this dynamically, but the use case is definitely clear. Implementing something like this would likely require some fairly deep groundwork that isn’t in place or planned right now.
For now, the count-in / pre-song voice cue approach is probably the most reliable workaround.
Hope that helps!