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

1. 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
2. For explicit volume/tone/recording requests, call the requested endpoint directly and handle its response. Use GET /audio for device diagnostics, not routine preflight. Relative volume changes still need GET /audio/volume.
3. Execute the appropriate API call
4. 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 /audio returns "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