react-native-voice-activatorThe on-device wake word for React Native voice agents

Listen for your phrase entirely on the device, then hand a hot mic to ElevenLabs, LiveKit or OpenAI Realtime. Or run the whole turn offline and never touch a network.

Get Started

Conversation Session API

View on GitHub

Any phrase, no training

wakePhrase 'hey acme' and you are done. No console, no dataset, no model to build. No cloud, no API key, no per-use cost.

The part voice agents are missing

Every speech-to-speech platform assumes the mic is already open. LiveKit ships on-device wake word for Python, Rust and Swift, not React Native. Deepgram says it does not do wake word at all.

Use it as a mic gate

addWakeWordListener fires, you hand the hot mic to whatever voice stack you already run. The handoff is where this package's job ends.

Or let it drive the turn

Opt in to STT and TTS providers and it runs the full listen, transcribe, your AI, speak loop with barge-in. Worth it offline; redundant when your vendor does turn detection server-side.

React hooks and an Expo plugin

useWakeWord() and useVoiceSession(), plus automatic native configuration, permissions, background modes and manifest entries.

New Architecture only

React Native 0.86+ and Expo SDK 57+, built as a TurboModule. No old-architecture bridge, which is where most wake-word packages are stuck.

What It Does

Quick Install

npm install react-native-voice-activator

Continue with Getting Started for platform setup (bare React Native or Expo), then Conversation Session for the full session API. Upgrading from an earlier release? See Upgrading.

Supports: React Native 0.86+ · Expo SDK 57+ · iOS · Android Expo Go is NOT supported. Use expo prebuild or EAS Build.

Built by Prompt Digital Agency · hello@prompt-digital.agency