Jarvis Physical V1

A desktop voice assistant with a physical OLED face built using Python and Arduino UNO .

The assistant listens for voice commands, transcribes speech locally using Whisper, performs system automation, answers questions using a local LLM, and displays real-time expressions on an OLED display.

Created by Divyanshu Verma , a Class 9 student .

---

Features

• Wake word activation
• Offline speech recognition using Faster Whisper
• Voice Activity Detection using Silero VAD
• Text-to-Speech responses
• Physical animated OLED face
• System automation
• Local AI conversation using Ollama
• Serial communication between Python and Arduino
• Command execution
• Continuous conversation mode

---

Hardware

• Arduino UNO
• 0.96" SSD1306 OLED Display (I2C)
• USB Cable
• Computer running Python

---

Software Stack

Python

• faster-whisper
• silero-vad
• sounddevice
• numpy
• pyttsx3
• requests
• pyserial

Arduino

• Adafruit GFX Library
• Adafruit SSD1306 Library
• Wire Library

---

OLED Expressions

The OLED display changes its facial expression depending on Jarvis' current state.

State	Expression
Idle	Calm face
Listening	Waiting for speech
Thinking	Animated processing face
Speaking	Talking animation

---

Voice Commands

Examples include:

• Open Notepad
• Open Browser
• Open GitHub
• Open YouTube
• Open VS Code
• Open Linux (WSL)
• Shutdown Computer
• Cancel Shutdown

Any command that is not recognized is automatically sent to the local AI model.

---

Project Structure

Jarvis/
│
├── jarvis1.1.py
├── arduino/
│   └── jarvis_face.ino
├── requirements.txt
├── LICENSE
├── README.md
└── .gitignore

---

Installation

Clone the repository

git clone https://github.com/Python-devloper-student/Jarvis-Physical-V1.git

Install dependencies

pip install -r requirements.txt

Upload the Arduino sketch to the Arduino UNO.

Start Ollama and make sure your custom model is available.

Update the COM port if necessary.

Run

python jarvis.py

---

How It Works

1. Jarvis waits for the wake word.
2. Voice is detected using Silero VAD.
3. Audio is transcribed locally using Faster Whisper.
4. If the command matches a predefined action, it is executed.
5. Otherwise, the request is sent to a local LLM through Ollama.
6. Responses are spoken aloud while the OLED face animates according to Jarvis' state.

---

Future Improvements

• ESP32 wireless version
• Battery powered hardware
• Camera integration
• Face recognition
• Home automation
• Mobile control
• GUI dashboard
• Memory system
• Vision capabilities

---

License

This project is licensed under the MIT License.

You are free to use, modify, and distribute this project, but proper credit to the original author is appreciated.

---

Author

Divyanshu Verma a class 9 student from india

GitHub: https://github.com/Python-devloper-student

Created as a personal learning project to explore embedded systems, AI, computer vision, and desktop automation.