EchoMind AI - Voice AI Agent with RAG

EchoMind AI is a functional Voice AI Agent that can "listen" to a user's question, "read" from a provided knowledge base (PDF/Text), and "speak" back the answer. The system prioritizes low latency to ensure the conversation feels natural.

Features

• STT (Speech-to-Text): Converts live microphone input to text using OpenAI Whisper.
• RAG Engine: Indexes documents (PDF/Text) and retrieves relevant context using LangChain and FAISS.
• Orchestration: Handles the flow and prompt grounding with GPT-4o-mini.
• TTS (Text-to-Speech): Converts the LLM's response back to audio using OpenAI TTS.
• MVC Architecture: Clean separation of concerns for maintainability.
• Streamlit UI: A premium, responsive dashboard for a great user experience.

Tech Stack

• Frontend: Streamlit
• LLM/STT/TTS: OpenAI API
• Vector Store: FAISS
• Retrieval: LangChain

Installation

1. Clone the repository:

   git clone https://github.com/Rahulgupta7777/Voice_Agent.git
   cd Voice_Agent

2. Install dependencies:

   pip install -r requirements.txt

3. Setup environment variables: Create a .env file in the root directory and add your OpenAI API key:

   DEEPGRAM_API_KEY-your_deepgram_api_key_here

Use

1. Run the application:

   streamlit run app.py

2. Upload documents: Use the sidebar to upload PDF or TXT files to build your knowledge base.

3. Interact: Hold the recording button to ask a question via voice, or use the text input.

Project Structure

• models/: RAG, Voice, and LLM core logic.
• controllers/: Orchestration logic.
• app.py: Streamlit interface.
• data/: Temporary storage for indexed documents