🤖 JARVIS AI Assistant (WIP 🚧)

JARVIS is a modular AI voice assistant built with Python that can listen, understand, respond, and perform basic actions using mostly free and local tools.

⚠️ This project is currently under active development.

---

🚀 Features (Current)

• 🎤 Voice input from microphone
• 🧠 Speech-to-text using faster-whisper (offline)
• 🤖 AI responses via Ollama (Llama3 local model)
• 🔊 Text-to-speech using pyttsx3 (offline)
• 🖼️ Image generation (Pollinations API, no API key)
• 🔁 Continuous conversation loop
• 🧩 Basic command detection (chat, image generation, shutdown)

---

🧠 System Flow

🎤 Voice → 🧠 STT → 🤖 LLM → 🔊 TTS → 🔁 Loop

---

🧱 Project Structure

Jarvis/
│
├── main.py # Entry point
│
├── core/
│ ├── listen.py # Microphone input
│ ├── transcribe.py # Speech → Text
│ ├── brain.py # LLM processing
│ ├── speak.py # Text → Speech
│
├── features/
│ ├── image_gen.py # Image generation
│
├── utils/
│ ├── config.py # Environment config
│
├── images/ # Generated images
├── .env
├── requirements.txt
└── README.md

---

⚙️ Tech Stack

• Python 3.11+
• faster-whisper
• pyttsx3
• Ollama (Llama3)
• sounddevice + scipy
• requests

---

▶️ How to Run

Follow these steps to run the JARVIS assistant locally.

---

1. Clone the Repository

git clone https://github.com/SifatNycto/Jarvis.git  
cd Jarvis

---

2. Create & Activate Virtual Environment

python -m venv .venv

Windows:

.venv\Scripts\activate

---

3. Install Dependencies

pip install -r requirements.txt

---

4. Install Ollama (Local AI Brain)

Download and install Ollama from:
👉 Ollama

Verify installation:

ollama --version

---

5. Download LLM Model

ollama pull llama3

---

6. Start Ollama Server

ollama serve

IMPORTANT: Keep this terminal running. This is the AI backend.

---

7. Run the Assistant

Open a NEW terminal (do not close the Ollama terminal):

python main.py

---

8. Use the Assistant

You can talk normally to the assistant.

Examples:

• "how are you"
• "what is python"
• "tell me a joke"
• "shutdown"

---

🖼️ Image Generation (Important)

To generate images, your command MUST include keywords like:

• "generate image"
• "generate an image"

Examples:

• "generate image of a cyberpunk city"
• "generate an image of a dragon"

If you do not include these words, the assistant will treat it as a normal conversation.

---

🛑 Stopping the System

Stop the assistant:

Ctrl + C

Stop Ollama:

taskkill /IM ollama.exe /F

---

🔥 Planned Improvements

• Better intent detection (LLM-based)
• Faster response pipeline
• Wake word support ("Hey Jarvis")
• Tool system (open apps, search web)
• GUI interface

---

🏁 Status

🚧 Work in progress — features and structure may change.

---

📌 Note

This project focuses on learning and building a real-world AI system using free/local tools with clean modular design.