A toggle would be fine I guess, but I would personally never use it. I like always having all information readily and immediately available in the corner instead of having to quickly look between speech bubbles strewn all over my screen. Then you gotta keep in mind how long the speech bubbles will be viewable, or what happens if someone speaks and they're not in your current field of view, and so on. Just...way too many ways of you missing something someone stated.

If there are people who can handle that mess, more power to em but personally I'm happy with just a chatbox.