Enable Voice Interaction

This guide walks you through enabling the voice pipeline in CoreLayer.

Prerequisites

Open Settings → Voice and toggle Enable Voice Pipeline.

Enable wake word detection to activate Jarvis by voice:

{
  "voice": {
    "wakeWord": {
      "enabled": true,
      "keyword": "jarvis"
    }
  }
}

The wake word runs locally using Picovoice Porcupine — no cloud processing.

Provider	Setup
Web Speech API	No setup needed — works out of the box
Groq Whisper	Requires Groq API key in Settings → Models

Web Speech API is recommended for getting started. Groq Whisper provides higher accuracy.

Say the wake word ("Jarvis") or click the microphone button in the Control Center. Then speak your request:

What tasks do I have for today?

Jarvis will process your speech, generate a response, and speak it back using TTS.

When voice is active, a floating overlay shows:

You can interrupt Jarvis mid-response by speaking again. The pipeline will stop the current response and process your new input.

Wake word not detected: Check microphone permissions and reduce background noise
Speech not recognized: Try speaking more clearly or switch to Groq Whisper
No audio output: Check system audio and TTS configuration