Since you can't get the dimensions of text right now, you could use another system to define how big a bubble you need. Like specify the bubble size at the start of the speech bubble text, as in "100|Hi there, this is a size 100 speech bubble" or "50|Size 50 bubble!" then use the gettoken system expression to get the size and bubble text. Or if you're using TimeLine, specify the size in a timeline parameter. And so on.
Whatever you do, don't use pre-drawn bubbles in animation frames. That will use loads of VRAM, and you'll quickly run out if you use a lot of bubbles. It'd be better to use one fixed size of bubble and fit your text inside of that, rather than pre-drawn bubbles.