Clarification on Live Stream Interruption, Chunked Script & Real-Time Interaction

1.Does the streams/stop endpoint fully stop a stream mid-speech, even when a talk request is still being processed?

2.Can we reuse the same stream ID after calling streams/stop to continue with a new chunk or response?

3.Is it safe and supported to send a series of short talk requests with small script chunks (30–50 characters) to simulate natural pacing?

Are there any rate limits or best practices for:

4.Sending multiple talk chunks rapidly?

5.Calling stop and talk in quick succession?

6.Are there any recommended strategies for syncing these chunks smoothly with real-time user mic detection and avatar responsiveness?