Voice channels run on LiveKit's WebRTC stack — real low-latency audio, not a bolted-on afterthought. Your input mode and device choices live in your own user settings, so they carry across every Outpost server you belong to.
Voice activity detection (VAD) transmits automatically whenever it detects you speaking — lowest friction, best for a quiet room. Push-to-talk only transmits while you hold a configured key — better in noisy environments or if you want explicit control over when you're heard. Both are available from your user audio settings; switch anytime without leaving a call.
Pick input (microphone) and output (speakers/headphones) devices independently in settings — useful if you're running a USB mic and Bluetooth headphones simultaneously, or need to switch outputs without interrupting a call.
Mute stops your own microphone from transmitting. Deafen goes further and stops incoming audio too (which also mutes you, since there's no point transmitting into a channel you can't hear). Both are available as quick toggles in the persistent voice control bar while you're connected — no need to open settings mid-call.
Your voice connection is tracked independently of which text channel you're viewing — browsing other channels while connected to voice won't disconnect you. The persistent voice bar stays visible with quick mute/deafen access regardless of where you navigate in the client.
