Add README.md for project documentation and update Dockerfile to include it.

Dockerfile

@@ -10,7 +10,7 @@ RUN apt-get update && apt-get install -y \
 
 RUN pip3 install uv
 
-COPY pyproject.toml uv.lock ./
+COPY pyproject.toml uv.lock README.md ./
 RUN uv sync --frozen --no-dev
 
 COPY src/ ./src/

README.md

@@ -0,0 +1,92 @@
+---
+title: Open Voice Agent
+emoji: 🎤
+colorFrom: blue
+colorTo: purple
+sdk: docker
+app_port: 8501
+pinned: false
+---
+
+# Open Voice Agent
+
+Real-time AI voice conversation application powered by LiveKit Agents, Moonshine STT, and Pocket TTS.
+
+## Features
+
+- **Streaming Speech-to-Text**: Moonshine (HuggingFace transformers)
+- **LLM Integration**: HuggingFace models or NVIDIA API via LangGraph
+- **Text-to-Speech**: Pocket TTS (Kyutai) with local inference
+- **Voice Activity Detection**: Silero VAD
+- **Web Interface**: Streamlit-based UI
+
+## Setup
+
+### Local Development
+
+1. Install dependencies:
+```bash
+uv sync
+source .venv/bin/activate
+```
+
+2. Configure environment:
+```bash
+cp .env.example .env
+# Edit .env with your API keys
+```
+
+3. Run the application:
+```bash
+# Terminal 1: Start LiveKit agent
+uv run src/agent/agent.py start
+
+# Terminal 2: Start Streamlit UI
+streamlit run src/streamlit_app.py
+```
+
+### Docker
+
+```bash
+docker build -t open-voice-agent .
+docker run -p 8501:8501 --env-file .env open-voice-agent
+```
+
+## Environment Variables
+
+### Required
+
+- `LIVEKIT_URL`: WebSocket URL for LiveKit server (wss://...)
+- `LIVEKIT_API_KEY`: LiveKit API key
+- `LIVEKIT_API_SECRET`: LiveKit API secret
+
+### LLM Provider (choose one)
+
+**HuggingFace** (local inference):
+```bash
+LLM_PROVIDER=huggingface
+HUGGINGFACE_MODEL_ID=Qwen/Qwen2.5-3B-Instruct
+HUGGINGFACE_DEVICE=cuda  # or 'cpu' or leave empty for auto
+HF_TOKEN=hf_xxx  # optional, for private models
+```
+
+**NVIDIA** (API-based):
+```bash
+LLM_PROVIDER=nvidia
+NVIDIA_API_KEY=nvapi-xxx
+NVIDIA_MODEL=meta/llama-3.1-8b-instruct
+```
+
+### Optional
+
+See `.env.example` for all available configuration options.
+
+## Requirements
+
+- Python >= 3.12, < 3.13
+- LiveKit server (cloud or self-hosted)
+- NVIDIA API key OR sufficient compute for local LLM inference
+
+## License
+
+Apache 2.0

src/agent/graph.py

@@ -45,8 +45,3 @@ def create_graph():
     workflow.add_edge(START, "agent")
     workflow.add_edge("agent", END)
     return workflow.compile()
-
-graph = create_graph()
-
-for msg in graph.stream({"messages": [{"role": "user", "content": "Hello, how are you?"}]}, stream_mode="messages"):
-    print(msg)

start.sh

@@ -4,5 +4,8 @@ set -e
 echo "Starting LiveKit agent..."
 uv run src/agent/agent.py start &
 
+echo "Waiting 30 seconds before starting Streamlit..."
+sleep 30
+
 echo "Starting Streamlit app..."
 exec streamlit run src/streamlit_app.py --server.port=8501 --server.address=0.0.0.0