Skip to content

samarth2703/BuddyAI

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation


🌟 Buddy AI — Your Personal Desktop AI Agent (Offline + Voice Controlled)

A fully offline-capable, voice-controlled AI assistant that can think, speak, search the internet, and control your Windows system — all running locally on your PC.

Buddy AI is lightweight, fast, privacy-friendly, and customizable.
_ made by - Samarth!!!

🧠 Offline AI Engine

🎤 Voice Recognition

🗣️ Text-to-Speech

🌐 Online Search

🪟 Windows App Controller

🤖 Built on Local LLM (Ollama)


🚀 Features

🧠 1. Offline AI Chat (Local LLM)

Buddy uses Ollama with the model:

llama3.2:1b

This model gives:

  • Low RAM usage
  • Fast responses
  • No internet required
  • Zero data sharing
  • Works even without GPU

Perfect for offline personal assistants.


🎤 2. Voice Input (Speech-to-Text)

Tools Used:Vosk (models/vosk-model-small-en-in-0.4) ✔ sounddevice

Features:

  • Always-on passive listening
  • Accurate Indian English recognition
  • Noise-resistant
  • Lightweight (~36 MB model)

Speak naturally and Buddy listens instantly.


🗣️ 3. Voice Output (Text-to-Speech)

Tools Used:Piper TTS (Fast, GPU/CPU optimized) ✔ playsound

Features:

  • Natural human-like voice
  • Very low latency
  • Completely offline
  • Supports many voice models

Buddy sounds real and responsive.


🌐 4. Online Search (DuckDuckGo API)

When user says:

"search <query>"
"google <query>"

Buddy performs online search using:

  • DuckDuckGo Instant Answer API
  • Returns clean text results
  • Works even on low bandwidth

🪟 5. Windows App & System Controller

Buddy can open, launch, or control your PC.

✔ Open ANY PC App

Examples:

"open chrome"
"open spotify"
"open vs code"
"open epic games"
"open calculator"

Even non-typical apps:

"open bluestacks"
"open obs studio"

✔ Control System

shutdown
restart
sleep

✔ Open Websites

open youtube
open google
open github

💻 6. Interactive Desktop GUI

Tools Used:tkinterScrolledText

GUI includes:

  • Real-time chat viewer
  • Live voice logs
  • Clean, minimal interface
  • Allows seamless voice + text interaction

🔧 Tech Stack Overview

Component Tool / Model Used
Local LLM Ollama + llama3.2:1b
Speech Recognition Vosk STT
Audio Input SoundDevice
Text-to-Speech Piper TTS
Voice Playback Playsound
Online Search DuckDuckGo Instant API
GUI Framework Tkinter
Windows Control Python subprocess, os, webbrowser
Python Version 3.10+

⚙️ How Buddy Works (Flow Diagram)

🎤 User Speaks →

🧠 Vosk STT Converts Speech →

🤖 Buddy Processes Intent

→ if command → Windows controller → if search → DuckDuckGo API → if chat → Local LLM

🗣️ Piper TTS Speaks Reply →

👤 GUI Displays Response

Everything happens smoothly and in real-time.


🏎️ Performance (8GB RAM Compatible)

Buddy is optimized for low-end devices:

  • Uses only 300–500MB RAM
  • Runs fully local LLM
  • No GPU required
  • Piper TTS runs in <0.2 sec
  • Vosk STT uses <50MB
  • Extremely fast and responsive

Perfect for:

  • Normal laptops
  • Old desktops
  • Portable installations

🔐 Privacy & Security

✔ Completely offline ✔ No server ✔ No cloud storage ✔ No data export ✔ Mic audio stays on your PC

Buddy is your truly private AI.


🌟 Why Buddy is Different

  • Doesn’t rely on cloud APIs
  • Extremely fast even on low hardware
  • Real Windows OS control
  • Offline + online hybrid
  • Modular code — easy to extend
  • Supports custom TTS voices
  • Works like Jarvis but locally
  • You can upgrade model anytime

❤️ Credits

  • Local LLM — Ollama
  • Offline Speech Recognition — Vosk
  • Text-to-Speech — Piper
  • GUI & Control — Python
  • Online Search — DuckDuckGo

🎉 Final Note

Buddy is not just a project — it’s a fully personal AI system designed to work on normal hardware with absolute privacy. You can keep extending it, upgrade models, add avatars, or turn it into a full desktop AI.


About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages