Obsidian Voice Agent

Obsidian Voice Agent

Voice and text interface for querying your Obsidian knowledge base using AI.

Features

• Voice Conversations: Real-time voice chat powered by OpenAI Realtime API (~300ms latency)
• Text Chat: Traditional chat interface for typed queries
• Knowledge Base Search: Full-text search across your Obsidian vault
• Real-time File References: See which documents the agent references as it speaks

Prerequisites

• Python 3.11+
• Node.js 20+
• uv (Python package manager)
• pnpm (Node package manager)
• LiveKit Cloud account (free tier available)
• OpenAI API key (with Realtime API access)

Quick Start

1. Clone and Install

git clone https://github.com/coleam00/obsidian-voice-agent.git
cd obsidian-voice-agent

# Backend
cd backend
uv sync

# Frontend
cd ../frontend
pnpm install

2. Configure Environment

Create `backend/.env`:

OBSIDIAN_VAULT_PATH=/path/to/your/vault
OPENAI_API_KEY=sk-...
LIVEKIT_API_KEY=your-api-key
LIVEKIT_API_SECRET=your-api-secret
LIVEKIT_URL=wss://your-project.livekit.cloud

Create `frontend/.env`:

VITE_LIVEKIT_URL=wss://your-project.livekit.cloud

3. Run

# Terminal 1: Start API server
cd backend
uv run uvicorn src.api.routes:app --port 8000

# Terminal 2: Start voice agent
cd backend
uv run python -m src.agents.voice_agent dev

# Terminal 3: Start frontend
cd frontend
pnpm dev

Open http://localhost:5173 and toggle between Text/Voice modes.

Architecture

┌─────────────────────────────────────────────┐
│              React Frontend                  │
│        (Text Chat + Voice UI)               │
└──────────────┬─────────────────┬────────────┘
               │                 │
         REST API          LiveKit Room
               │                 │
┌──────────────▼──────┐  ┌──────▼──────────────┐
│   Text Agent        │  │   Voice Agent       │
│   (Pydantic