feat(voice): add requireHeadphones config to gate voice on audio output device

[blu3dot]

Summary

New voice.requireHeadphones setting that gates voice playback on the audio output device type. When enabled, voice is skipped through built-in laptop speakers but plays normally through headphones, AirPods, USB DACs, HDMI audio, and other external devices.

Problem

Users in shared environments (offices, cafes, meetings) have no way to prevent voice from blasting through laptop speakers. The only option is disabling voice entirely, losing the feature when headphones are connected.

Solution

Config (settings.json):

{
  "voice": {
    "requireHeadphones": false
  }
}

Default false — opt-in, non-breaking. Uses === true so missing key defaults to OFF.

Detection — queries system_profiler SPAudioDataType -json and checks the CoreAudio Transport type of the default output device:

Transport	Device Type	Voice Plays?
coreaudio_device_type_builtin	MacBook speakers	No
coreaudio_device_type_bluetooth	AirPods, BT headphones	Yes
coreaudio_device_type_usb	USB DAC, dock audio	Yes
coreaudio_device_type_hdmi	External monitor	Yes
coreaudio_device_type_airplay	HomePod, Apple TV	Yes
All other transport types	External audio	Yes

Key design decisions:

• 30-second cache — system_profiler takes 140-250ms; cache prevents repeated slow calls during burst scenarios (e.g., 10 notifications during a build)
• 3-second timeout — prevents hangs if system_profiler stalls
• Fail-open — if detection fails or times out, voice plays anyway (convenience feature, not security gate)
• Desktop notifications always display regardless of headphone state
• Uses -json flag for reliable machine-parseable output (no text scraping)

macOS-only — system_profiler is macOS-specific. Linux equivalent is out of scope (see #855, #872 for cross-platform audio work).

References #855

Test plan

• Connect Bluetooth headphones → requireHeadphones: true → voice plays
• Disconnect headphones (Built-in) → voice skipped, log shows 🔇
• requireHeadphones: false (default) → voice plays regardless of output device
• /health endpoint shows require_headphones field
• Detection failure → voice plays (fail-open)
• Config key absent → voice plays (opt-in default OFF)

Post-Deploy Monitoring & Validation

No additional operational monitoring required: opt-in feature defaulting to OFF, zero behavior change for existing users.

---

🤖 Generated with Claude Code

[blu3dot]

Closing to test locally before resubmitting.