VoiceRAG

• It's a Voice RAG project built with LiveKit agent. It uses :

🟢 Silero VAD to detect when I start speaking

🗣️ AssemblyAI to convert my speech to text

📄 ChromaDB for vector search across uploaded documents

🧠 LLM (Gemini/OpenAI) to generate intelligent responses

🗣️ Cartesia TTS to speak the reply out loud

🧬 And finally… Cartesia Voice Cloning so the reply comes in my own voice

❤️ Livekit for orchestration

FlowChart

Command to run after cloning and installing the requirements

1. If you want to connect and check in liveKitPlayground

 uv run voiceRAG.py dev 

2. If you want to check in the console(cmd)

 uv run voiceRAG.py console 

3. If you want to check in the production mode

 uv run voiceRAG.py start 

Note: In the dev and start modes, your agent connects to LiveKit Cloud and joins rooms:

• dev mode: Run your agent in development mode for testing and debugging.

• start mode: Run your agent in production mode.