Press Start call and allow the microphone. The agent greets you, then just talk. It replies when you pause.
Timings under each reply: STT = speech to text, LLM = reply generation, TTS = voice synthesis, total = end of your speech until the agent starts talking.