Browser audio that starts on purpose

gesture-audio provides the small, composable infrastructure behind dependable interactive audio. It starts Web Audio only after a real gesture, routes continuous Tone.js signals through a caller-named bus graph, plays Howler sprites, and applies persisted preferences without owning an application's cue naming or storage model.

The central guarantee is intentionally narrow: a successful start means both Tone.start() and the application's bootstrap completed. If either fails, the next gesture can retry rather than leaving the application in a half-ready state.

What it provides

• A typed Tone.js bus graph with a master limiter, independent mute layers, and timed ducking.
• A Howler sprite resolver that validates maps and keeps active loops in sync with live mix changes.
• A bridge from a small asynchronous preferences store to both engines.
• Node-only sprite verification from the separate build-tools entry point.

What stays in your application

Cue names, gameplay policy, persistence implementation, and product-specific audio decisions are deliberately local. Start with the quick start then follow the complete integration guide. The architecture records the invariants that make those boundaries reliable.