Real-time audio director for tabletop RPGs.

DMBox listens to your D&D session, recognizes scene mood and combat phases, and plays appropriate music + sound effects automatically. The DM gets a 2×2 quadrant of one-click picks for everything the AI isn't sure about. Players hear what the DM commits to — never the spoilers.

See pricing Read the code

How it works

Four layers cooperate per session: a browser-side Web Speech ASR feeds transcripts to a trigger matcher; a per-tier LLM (or a deterministic rules picker on Free) infers scene phase and recommends BGM + SFX from a license-cleared library; a slot mixer enforces buses and ducking; the cockpit surfaces the choices the AI made and the ones it wants the DM to pick.

Players see a separate, spoiler-free view of the same session. No DM buttons, no preview audio, no future-pick teasers. Plays the audio the DM committed to, nothing more.

What's bundled

~30 starter tracks — CC0 or CC-BY-only, attribution shipped per track.
Scene-profile templates for fantasy, horror, and sci-fi genres.
Discord and Slack webhook integrations for play-by-post tables.
Magic-link auth, per-session join codes, audit trail.
Open-source under BSL 1.1; converts to Apache 2.0 four years after each release.

Status

Pre-launch beta. Public launch within the next 30 days. Sign up for early access on the GitHub repo — open an issue with "[beta]" in the title and we'll add you when the cockpit is live.