Built as part of the Mem0 AI — MLOps and AI Infra Internship Assignment

A voice-controlled AI agent that accepts audio input, classifies intent, executes local tools, and displays results in a clean UI — all powered by Groq's free API.

🎥 Video Demo: [YouTube Link Here]
📝 Article: [Dev.to/Medium Article Link Here]

• 🎤 Dual Audio Input — Record from microphone OR upload .wav/.mp3 files
• 🔊 Speech-to-Text — Powered by Groq Whisper Large V3
• 🧠 Intent Detection — LLaMA3-70B classifies user commands into 4 intents
• ⚡ Tool Execution — Creates files, writes code, summarizes text, or chats
• 🧩 Session Memory — Tracks all actions taken during the session
• 🖥️ Clean Gradio UI — Shows transcription, intent, action, and output

User speaks
    ↓
Audio file (.wav/.mp3)
    ↓
[stt.py] Groq Whisper → Transcribed Text
    ↓
[intent.py] LLaMA3-70B → Intent Classification
    ↓
[tools.py] Tool Router
    ├── create_file  → Creates file in output/
    ├── write_code   → Generates code, saves to output/
    ├── summarize    → Summarizes text, saves to output/summary.txt
    └── general_chat → LLM conversation
    ↓
[app.py] Gradio UI → Displays all results
    ↓
Session History (In-memory)

This project uses Groq's API for both Speech-to-Text and LLM inference instead of running models locally. This was chosen because:

1. Groq provides free tier access
2. It achieves ultra-fast inference (faster than most local setups)
3. It makes the project hardware-agnostic — runs on any machine
4. Whisper Large V3 and LLaMA3-70B are available on Groq for free

• Python 3.9 or higher installed
• A free Groq account

git clone https://github.com/YOUR_USERNAME/voice-ai-agent.git
cd voice-ai-agent

pip install -r requirements.txt

1. Go to console.groq.com
2. Sign up for free
3. Click API Keys → Create API Key
4. Open the .env file and replace your_groq_api_key_here with your actual key:

GROQ_API_KEY=gsk_xxxxxxxxxxxxxxxxxxxxx

python app.py

Then open your browser and go to: http://localhost:7860

voice-ai-agent/
├── app.py              ← Main Gradio UI application
├── stt.py              ← Speech-to-Text (Groq Whisper)
├── intent.py           ← Intent classification (LLaMA3)
├── tools.py            ← Tool execution (files, code, summarize, chat)
├── config.py           ← Loads .env file
├── requirements.txt    ← Python dependencies
├── .env                ← Your API key (never commit this!)
├── .gitignore          ← Excludes .env from git
└── output/             ← All generated files go here (auto-created)

• ✅ Session Memory — All actions tracked throughout the session
• ✅ Compound Commands — Handles multiple intents in one command
• ✅ Graceful Degradation — Clear error messages for all failure cases
• ✅ Safety Constraint — ALL file operations restricted to output/ folder

Built for the Mem0 AI MLOps & AI Infra Internship assignment. Inspired by the concept of persistent memory in AI agents — a core problem that Mem0 is solving.