autonomous-ai Audio Control
Quick Start
Control the device's speaker and microphone hardware directly. Use this for volume adjustments, test tones, and raw audio recording. This is LOW-LEVEL hardware control only.
Workflow
- Determine the user's audio hardware need:
- Volume adjustment -> use
POST /audio/volume - Check current volume -> use
GET /audio/volume - Diagnostics / test -> use
POST /audio/play-tone - Raw recording -> use
POST /audio/record
- Volume adjustment -> use
- For explicit volume/tone/recording requests, call the requested endpoint directly and handle its response. Use
GET /audiofor device diagnostics, not routine preflight. Relative volume changes still needGET /audio/volume. - Execute the appropriate API call
- Confirm the action to the user
Examples
Input: "Louder please" / "Turn it up"
Output: Check current volume with GET /audio/volume, then increase by ~15 with POST /audio/volume. Confirm: "Volume set to 85%."
Input: "Set volume to 50%"
Output: Call POST /audio/volume with {"volume": 50}. Confirm: "Volume set to 50%."
Input: "Mute" / "Too loud"
Output: Call POST /audio/volume with {"volume": 0}. Confirm: "Muted."
Input: "I can't hear you"
Output: Check current volume with GET /audio/volume, then increase it. Confirm with the new level.
Input: "Say something" / "Tell me a joke" Output: Do NOT use this skill. Just reply normally — your voice pipeline handles TTS automatically.
Tools
Use Bash with curl to call the HTTP API at http://127.0.0.1:5001.
Check audio devices
curl -s http://127.0.0.1:5001/audio
Response:
{
"output_device": 0,
"input_device": 1,
"available": true
}
Set volume
curl -s -X POST http://127.0.0.1:5001/audio/volume \
-H "Content-Type: application/json" \
-d '{"volume": 70}'
Volume range: 0 (mute) to 100 (max).
Get current volume
curl -s http://127.0.0.1:5001/audio/volume
Response: {"control": "Speaker", "volume": 70}
Play test tone
curl -s -X POST "http://127.0.0.1:5001/audio/play-tone?frequency=440&duration_ms=500"
Plays a sine wave. Use for audio testing only. Keep it short (< 1 second).
Record audio
record_dir=$(mktemp -d /tmp/autonomous-recording.XXXXXX) &&
curl --fail --silent --show-error -X POST \
"http://127.0.0.1:5001/audio/record?duration_ms=3000" \
--output "$record_dir/recording.wav" &&
printf 'Recording saved: %s\n' "$record_dir/recording.wav"
Records from the microphone and returns WAV bytes; save them to a unique file rather than printing binary data to the model. Adapt the duration to the user's request. Report the path only after the request succeeds; a failed request may leave an empty or partial file and is not a recording result. Do not re-record automatically after an uncertain outcome.
Error Handling
- If
GET /audioreturns"available": false, inform the user: "The speaker/microphone is not connected right now." - If the API is unreachable, inform the user: "I couldn't access the audio hardware. The service may be unavailable."
- If the user requests a volume outside 0-100, clamp to the valid range.
Rules
- Default volume is usually 70%. Adjust based on user preference.
- Audio = volume knob, raw mic recording, test beeps. No AI speech processing.
- Voice = AI-powered speech (TTS/STT). Uses Audio hardware underneath but is a separate skill.
- When the user says "I can't hear you" or "too loud", adjust volume via this skill.
- Do NOT use this skill for TTS or speech output — that is handled by the Voice skill and the automatic voice pipeline.
Output Template
[Audio] {action} — {details}
Examples:
[Audio] Volume set — 70%[Audio] Volume set — muted (0%)[Audio] Test tone played — 440Hz, 500ms[Audio] Recording captured — 3000ms