How parshxv/voice_agent works
Veena AI is a bilingual voice insurance agent: a customer speaks or types through the web interface, the backend transcribes microphone audio or accepts text, the conversational RAG agent retrieves policy context and generates a reply, and the voice service speaks it back. The README also describes an agent dashboard and optional Supabase integration, but those frontend modules are not present in the supplied tree. Edges follow shown calls or README-documented behavior; browser-to-server wiring and unsampled features are left unasserted where evidence is absent.
Main entry points: app.py Web and audio server · agentic_rag.py Conversation agent · index_documents.py Document indexing · voice_service.py Speech synthesis
Components 18
Web experience
- Customer interface
frontend/index.html - Agent dashboard
frontend
Conversation runtime
- Web and audio server
app.py - Conversation agent
agentic_rag.py - Chat memory
agentic_rag.py - Groq language model
Knowledge retrieval
- Policy retriever
agentic_rag.py - FAISS index
faiss_rag.index - Document indexing
index_documents.py - Policy documents
rag_docs - Ollama embeddings
Speech services
- Audio capture
app.py - Speech recognition
app.py - Speech synthesis
voice_service.py - Edge TTS
- Audio playback
voice_service.py
Other components
- Customer
- Supabase
Connections 16
- Customer → Customer interface: uses
README.md - Customer interface → Web and audio server: sends interaction
- Web and audio server → Audio capture: records audio
app.py - Audio capture → Speech recognition: transcribes
app.py - Web and audio server → Conversation agent: invokes
app.py - Conversation agent → Chat memory: maintains context
agentic_rag.py - Conversation agent → Policy retriever: calls tool
agentic_rag.py - Policy retriever → FAISS index: searches
agentic_rag.py - Document indexing → Policy documents: reads
index_documents.py - Document indexing → Ollama embeddings: embeds chunks
index_documents.py - Document indexing → FAISS index: writes index
index_documents.py - Conversation agent → Groq language model: generates response
agentic_rag.py - Web and audio server → Speech synthesis: requests speech
app.py - Speech synthesis → Edge TTS: generates audio
voice_service.py - Speech synthesis → Audio playback: plays audio
voice_service.py - Agent dashboard → Supabase: optional integration
README.md