Jarvis – Voice Controlled Desktop Assistant

Project Description

Jarvis is a Python-based voice-controlled personal assistant that uses Google's Speech Recognition API and the pyttsx3 text-to-speech engine to respond to user commands. It works on a wake-word system — when the user says "Jarvis", the assistant activates and listens for the next instruction. It can open websites, tell the current time, respond to basic conversational queries, and gracefully shut down when instructed. The system includes ambient noise handling, error handling, and continuous real-time microphone listening.

Key Features

• Wake-word activation using the keyword "Jarvis".
• Voice-controlled automation of websites (Google, YouTube, Facebook).
• Speaks current system time on request.
• Responds to conversational prompts like “How are you?”.
• Graceful shutdown when the user says “stop”.
• Error handling for ambient noise, silence, or recognition failures.

Applications

• Hands-free desktop control.
• Accessibility tool for users with mobility limitations.
• Voice automation for productivity.
• Customizable assistant for personal workflow automation.
• Educational project for learning speech recognition + Python automation.

Future Enhancements

• Add wake-word detection without constant microphone polling (Snowboy / Porcupine).
• Integrate with ChatGPT or LLMs for intelligent dialogue.
• Add system-level controls (volume, brightness, music playback, app launching).
• GUI dashboard showing logs and system activity.
• Advanced NLP instead of keyword matching.
• Offline speech recognition using Vosk / Whisper.

Source Code

import speech_recognition as sr
import pyttsx3
import webbrowser
import time

recognizer = sr.Recognizer()
engine = pyttsx3.init()

def speak(text):
    engine.say(text)
    engine.runAndWait()

def processCommand(command):
    command = command.lower()

    if 'open google' in command:
        speak("Opening Google.")
        webbrowser.open("http://www.google.com")

    elif 'open youtube' in command:
        speak("Opening YouTube.")
        webbrowser.open("https://www.youtube.com")

    elif 'open facebook' in command:
        speak("Opening Facebook.")
        webbrowser.open("https://facebook.com")

    elif 'time' in command:
        current_time = time.strftime("%I:%M %p")
        speak(f"The current time is {current_time}.")

    elif 'how are you' in command:
        speak("I'm doing great, thank you for asking!")

    elif 'stop' in command:
        speak("Goodbye! I am shutting down.")
        exit()

    else:
        speak("Sorry, I didn't understand that.")

while True:
    try:
        with sr.Microphone() as source:
            print("Listening for wake word...")
            recognizer.adjust_for_ambient_noise(source)
            audio = recognizer.listen(source, timeout=5, phrase_time_limit=3)

        word = recognizer.recognize_google(audio)
        print("Recognized:", word)

        if word.lower() == "jarvis":
            speak("Yes, how can I assist you?")
            with sr.Microphone() as source:
                recognizer.adjust_for_ambient_noise(source)
                audio = recognizer.listen(source, timeout=5, phrase_time_limit=5)
                command = recognizer.recognize_google(audio)
                processCommand(command)

        except Exception as e:
        print("Error:", e)