These values map to the same realtime WebSocket flow used by your existing app.
Select the LoomyLlama model the realtime websocket should use directly.
Select the microphone input and audio output device for this avatar page.
Select one of your cloned voices. Voice design is not used here.
Lower values are faster. Higher values are slower, but usually improve audio quality. It is always a tradeoff between speed and quality.
Connect to your Agent Brain record. After verification, the brain matter field is appended to your system prompt automatically. Manage your record in the popup.
This prompt is sent in the realtime conversational config before audio starts.
Save the current cloned-voice settings for this avatar page.
Manage your Agent Brain record.