Voice is configured in the Happy mobile app under Settings > Voice.

​
Happy Voice (default)

Happy Voice is built into the app and works out of the box — no additional setup or registration required. Just tap the microphone icon in any session to start a voice conversation. Happy Voice uses a voice gateway based on LiveKit with a STT → LLM → TTS pipeline. The hosted gateway is provided for all users by default.

​
Self-hosted customization

If you want full control over the voice pipeline — choosing your own STT, LLM, and TTS models, or keeping voice data on your own infrastructure — you can deploy your own Happy Voice gateway. Requirements:
  • A LiveKit server instance
  • API keys for your chosen providers (e.g., OpenAI for STT/LLM, Cartesia for TTS)
In the Happy app, go to Settings > Voice and configure:
  • Provider — Select “Happy Voice”
  • Gateway URL — Your self-hosted Happy Voice endpoint
  • Public Key — The authorization key for your voice server
Or set via environment variables: 
EXPO_PUBLIC_VOICE_PROVIDER=happy-voice
EXPO_PUBLIC_VOICE_BASE_URL=https://your-voice-server.com
EXPO_PUBLIC_VOICE_PUBLIC_KEY=your-public-key
See Self-Hosting for deployment instructions.

​
ElevenLabs (alternative)

ElevenLabs is a third-party voice AI service with high-quality speech synthesis. It requires a free or paid ElevenLabs account.
  1. Create an account at elevenlabs.io and get an Agent ID
  2. In the Happy app, go to Settings > Voice
  3. Select ElevenLabs as the provider
  4. Enter your Agent ID
Alternatively, set the Agent ID via environment variable: 
EXPO_PUBLIC_ELEVENLABS_AGENT_ID=your-agent-id