ConvoxAI

AI-powered call transcription, summarization, and analysis platform with RAG chatbot capabilities.

Overview

ConvoxAI is an intelligent call analysis system that transcribes audio files, generates contextual summaries with sentiment analysis, and provides an interactive AI chatbot for natural language querying. Built with FastAPI and React, powered by state-of-the-art language models.

Features

Feature	Description
Audio Transcription	Convert audio files to text using Faster Whisper (WAV, MP3, M4A, FLAC, OGG)
AI Summarization	Generate detailed summaries with sentiment analysis, key insights, and action items
RAG Chatbot	Query call history using natural language with context-aware responses
Secure Authentication	User management powered by Supabase Auth with JWT tokens
Cloud Storage	Persistent audio file storage with Supabase Storage
PDF Reports	Export professional call summary reports
Real-time Processing	Instant audio analysis and insights delivery

Tech Stack

Backend

Component	Technology
Web Framework	FastAPI (Python 3.12+)
LLM Orchestration	LangChain, LangGraph
Audio Processing	Faster Whisper, PyDub
Vector Database	Pinecone
Database	Supabase (PostgreSQL)
Authentication	Supabase Auth + JWT

Frontend

Component	Technology
UI Framework	React 18+ with TypeScript
Build Tool	Vite
Styling	Tailwind CSS 4
UI Components	Radix UI + shadcn/ui
HTTP Client	Axios

---

Project Structure

ConvoxAI/
├── Backend/                    # FastAPI Backend Service
│   ├── app/                    # API routes and endpoints
│   ├── core/                   # Business logic (chatbot, summarizer)
│   ├── utils/                  # Utilities (audio, embeddings, validation)
│   ├── database/               # SQL schemas and RLS policies
│   ├── config.py               # Centralized configuration
│   └── main.py                 # Application entry point
│
├── Frontend/                   # React Frontend Application
│   ├── src/
│   │   ├── components/         # React components (auth, ui, pages)
│   │   ├── contexts/           # Auth context provider
│   │   ├── lib/                # API client and utilities
│   │   └── types/              # TypeScript definitions
│   └── vite.config.ts          # Vite configuration
│
└── README.md

Detailed documentation:

• Backend/README.md — API reference, configuration, database schema
• Frontend/README.md — Component structure, features, deployment

---

Requirements

Requirement	Details
Python	3.12+
Node.js	18+
Supabase	Account for database, auth, and storage
Pinecone	Account for vector database
AI Provider	Google Gemini or Groq API key

---

Quick Start

Backend Setup

cd Backend
python3.12 -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
# Create .env with API keys (see Backend/README.md)
python main.py

Server runs at: http://localhost:8000

Frontend Setup

cd Frontend
npm install
# Create .env with Supabase config (see Frontend/README.md)
npm run dev

App runs at: http://localhost:5173

See Backend/README.md and Frontend/README.md for complete setup instructions.

---

Usage

1. Sign Up / Login — Create an account with email and password
2. Upload Audio — Drag and drop or select audio file (max 50 MB)
3. View Summary — See AI-generated summary with transcript, sentiment, and key points
4. Chat with AI — Ask natural language questions about your calls
5. Export — Download PDF reports of call summaries

Example Chat Queries

• "What were the main topics discussed?"
• "What was the overall sentiment of the conversation?"
• "What are the key action items mentioned?"

---

Configuration

Backend (Backend/config.py)

# Language Models
GEMINI_MODEL_NAME = "gemini-2.5-flash"
GROQ_MODEL_NAME = "qwen/qwen3-32b"

# Audio Transcription
WHISPER_MODEL_SIZE = "tiny"  # tiny | base | small | medium | large

# Embeddings
EMBEDDINGS_MODEL_NAME = "sentence-transformers/all-MiniLM-L6-v2"

Frontend (.env)

VITE_SUPABASE_URL=your_supabase_url
VITE_SUPABASE_ANON_KEY=your_supabase_key
VITE_API_BASE_URL=http://localhost:8000

---

API Documentation

Interactive API documentation available when backend is running:

• Swagger UI: http://localhost:8000/docs
• ReDoc: http://localhost:8000/redoc

See Backend/README.md for complete endpoint reference.

---

Contributing

1. Fork the repository
2. Create a feature branch: git checkout -b feature/your-feature
3. Commit changes: git commit -m "Add: description of changes"
4. Push to your fork: git push origin feature/your-feature
5. Open a Pull Request

---

License

MIT

---

Resources

Resource	Link
Backend Documentation	Backend/README.md
Frontend Documentation	Frontend/README.md
Issue Tracker	GitHub Issues
API Documentation	http://localhost:8000/docs