How parthivshah33/realtime-rag-voicebot works
Project type and purpose
• This is a Python‐based voice assistant (“voicebot”) proof of concept. It captures live audio, converts speech to text, processes it (optionally with retrieval-augmented generation), and returns spoken answers via text-to-speech. The “experiments” folder contains notebooks and scripts for exploring different STT/TTS engines, embedding generation, RAG, and LangChain memory.
File structure
• Top level:
• voicebot.py: main orchestration script
• logger.py: centralized logging
• .env + requirements.txt: configuration and dependencies
• utils/: core modules for STT, TTS, embeddings, RAG processing
• experiments/: exploratory notebooks and scripts for different speech/LLM pipelines
• knowledge documents/: static PDFs used as retrieval corpus
• Patterns: single-process pipeline, modular handlers for each stage (STT → LLM → TTS), external API reliance (Deepgram, OpenAI, etc.).
README for additional insights
• Basic setup steps: virtualenv, install deps, add API keys, run voicebot.py.