Voice chat

The table can talk over live voice in the lobby and during play, with no third-party service and no account. On a world hosted from an app, voice is on and runs browser to browser through the share tunnel. On a self-hosted server the administrator turns it on and picks the transport (see Voice chat setup). Either way, the controls you use are the same.

Joining voice

The Voice dock in the table header (and the lobby) joins the call. The first join asks for microphone permission: the browser's prompt, or the app's on Android. Once in, the dock shows who is speaking, your mute switch, and volume sliders for voice chat, narration and ambience. The Dungeon Master is never muted and always hears everyone.

Your microphone and speaker

Device settings, reachable from the dock, from Settings in the app, and from the table, hold the recording and playback choices: which microphone, a recording level with a live meter and a test, open mic or push to talk, the output device with a test tone (where the browser can route audio), and the three volume switches. A change applies live to a call already in progress.

The floor and raised hands

A campaign setting decides how turn-taking works in voice: ignored (everyone talks freely), shown (the default: whose turn it is shows in the dock, nothing is enforced), or enforced (only the current speaker and the Dungeon Master are audible). The raise-hand button beside the message box asks for the floor; it is a request, and the lead or Dungeon Master grants it.

Room rules

Optional rules, all per campaign and off by default, make the room feel like a room:

• Proximity. Characters hear each other within 30 feet on the battle map; a whisper carries 5 feet, a shout 120. Walls muffle.
• Downed go deaf. An unconscious character hears nothing until they are up.
• Side rooms. The Dungeon Master can pull one or two players into a breakout for a private scene, and bring them back.

A member the party lead has muted is silent on the call too and adds nothing to the transcript: on a relayed call they keep listening, on a device-to-device call they leave it. Where the server runs the full media server, these are enforced by the server. On a mesh call (an app-hosted world, or a server that chose mesh), gains are applied by each listener's own device and there is no speaking indicator, floor enforcement or breakout rooms.

Push-to-talk transcription

Different from voice chat: the microphone button beside the message box records a short clip, sends it to the server's speech-to-text service if one is configured, and shows the text for you to confirm before it is sent as your action. It needs a server with a speech-to-text backend: a Whisper service, or the built-in Whisper an administrator installs from the admin panel's Speech section. A device world has none until its owner sets one up there. On a Steam Deck or with a controller, L1 holds voice chat's push to talk.

When nobody can hear you