Adds code to have better control over the agent. Now, you can pause s…

[yepher]

See issue: #1118 - #1118

In my application case, I need to have the ability to "pause" the Multimodal agent so I can speak to participants in the channel without the LLM interrupting me. I also need to be able to resume the LLM interaction.

In addition to pausing the resume, I also need to be able to set the source participant the agent should be actively listening to. In my app there are multiple participants in the room.

Locally I am doing this and it seems to work:
update multimodal_agent.py

    def set_source_participant(self, participant_identity: str) -> None:
        """Set the source participant for the agent."""
        self._link_participant(participant_identity)

    def pause_listening(self) -> None:
        """Pause listening to the audio stream."""
        if self._read_micro_atask is not None:
            self._read_micro_atask.cancel()  # Stop the current task
            self._read_micro_atask = None
            logger.info("Paused listening to the audio stream.")

    def resume_listening(self) -> None:
        """Resume listening to the audio stream."""
        if self._subscribed_track is not None and self._read_micro_atask is None:
            self._read_micro_atask = asyncio.create_task(
                self._micro_task(self._subscribed_track)  # Restart the task
            )
            logger.info("Resumed listening to the audio stream.")

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.

---

Chris Wilson seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}