Skip to content

Voice bridge·

text → the voice persona's speaker from the text personastools/voice_bridge.py, 1 tool.

From the module docstring

Voice bridge — async text-input queue feeding the bidi voice agent.

tool does
voice_say Send a message to Scout's voice agent so it speaks/reacts out loud.

voice_say·

voice_say(text: str, importance: int = 1, source: str = 'manual') -> dict

Send a message to Scout's voice agent so it speaks/reacts out loud.

Use this from any other persona (telegram, thinker, shell) to make the voice agent react to something. The voice agent picks it up from a shared SQLite queue within ~2 seconds and decides how to respond (often with MOTION rather than words, per Scout's persona).

argument meaning
text What the voice agent should know/react to. Plain text.
importance 0=silent log, 1=normal info (default), 2=URGENT URGENT briefings are surfaced immediately mid-conversation.
source who is sending (telegram|thinker|agent|manual). Default manual.

Returns

dict with 'id' of the queued message and 'status'.

Examples

voice_say("the user just said hi over telegram", source="telegram")
voice_say("CRITICAL: battery at 12%, head home", importance=2, source="thinker")

source: tools/voice_bridge.py:126