Voice mode: choppy responses to requests

This one's kind of hard to give an exact repro, but pretty much, I started voice mode and made a request to create a session to investigate a telemetry issue. Voice mode would respond in the middle of the request or after I've completed the request, asking, "What do you want the new session to do? What do you want the new session to investigate? What do you want?" It finally got to the point where it actually did what I asked without further input from me. It's as if the model is only getting snippets of what I say, but my internet connection is pretty reasonable, and I don't have this issue with other voice-driven apps.