I actually really like Mark’s approach. Notating the explicit |—2—| has the advantage of “signaling loudly” the shift in metrical feel, and something like the “straight ♫” confirms it.
I’m not sure using different bars/time signatures in different parts is such a good idea. You don’t want to waste time in rehearsal when the conductor says ‘let’s start at the pick-up for bar x’, and half the orchestra has no pick-up bar at all.
I’d use either a duplet at the end of the 12/8, then a TS change to 4/4 in the next bar, or insert a change to 4/4 with a ¼ pickup right on the last beat (=the 10th quaver) of the 12/8, including a double bar for good measure (pun unintended). Like you apparently did in the example in → this thread:
But the voice part will have a pickup where the piano wouldn’t have anyway, no matter what meter changes I introduce. Also - in that case you still have 4 beats in both parts, it’s just the subdivision that changes. (So you can say stuff like “let’s start from the 3 of bar x”)
But isn’t this just for voice and piano? Wouldn’t both performers be seeing the score,
Yes, it’s just voice and piano, but also want to give the performers the option to read only from a vocal line (to save on paper and page flips). That’s another reason why I want it to make sense from the singer’s perspective, so keep it straight when it straight. If I’d write duplets, it would look like a mistake from the perspective of the singer who doesn’t know about the consideration about the piano.
I’m wondering if interchangeable time sigs would be helpful here. As in 4/4=12/8. It seems like that might signal the metrical tension to the singer. Hmm…