VoiceRx

AI-powered voice health assistant for patients who can't access doctors.

Tech Stack

• Vapi — Voice AI (call handling, speech-to-text, text-to-speech)
• Qdrant — Vector database for semantic FAQ retrieval
• Groq LLM — llama-3.1-8b-instant for synthesizing answers
• FastAPI — Webhook server for Vapi tool calls
• FastEmbed — Local embeddings (BAAI/bge-small-en-v1.5)

How It Works

Patient speaks
     │
     ▼
Vapi (speech-to-text)
     │
     ▼
FastAPI /search webhook
     │
     ▼
Qdrant RAG (top 3 semantically similar FAQs)
     │
     ▼
Groq LLM (synthesizes answer using retrieved context)
     │
     ▼
Vapi (text-to-speech)
     │
     ▼
Patient hears response

Architecture

Patient speaks → Vapi (STT) → Tool Call → FastAPI /search endpoint → FastEmbed (embed query) → Qdrant (semantic search, top 3 results) → Groq LLM (synthesize answer with context) → Vapi (TTS) → Patient hears response

Setup

1. Install dependencies

pip install fastapi uvicorn qdrant-client fastembed groq httpx python-dotenv

2. Set environment variables

Create a .env file in the project root:

QDRANT_URL=your_qdrant_cluster_url
QDRANT_API_KEY=your_qdrant_api_key
GROQ_API_KEY=your_groq_api_key

3. Ingest health FAQs into Qdrant

python ingest.py

4. Start the server

uvicorn main:app --reload --port 8000

The webhook will be available at http://localhost:8000/search.

---

Built for HackBLR 2026