SINK

A developer built SINK, an open-source JARVIS-style desktop assistant that runs locally using Python, PySide6, Faster-Whisper for speech recognition, and Qwen 3 served through LM Studio. The assistant executes voice commands to open applications, control the mouse and keyboard, type, scroll, manage tasks and reminders, and search YouTube, with text-to-speech handled by pyttsx3 and desktop automation by PyAutoGUI. The project is positioned as a privacy-preserving alternative to cloud-dependent assistants, keeping conversations and speech processing on the user's machine.

I built SINK, a JARVIS-style local AI desktop assistant that lets users control their PC using voice commands and interact with an AI assistant without relying entirely on cloud services. SINK can open applications, control the mouse and keyboard, scroll, type, manage tasks and reminders, answer questions, assist with studying, and interact with YouTube. It also has an animated robot face that reacts to conversations. The goal is simple: make everyday computer interactions faster, more natural, and hands-free. Instead of manually switching between applications and repeating routine actions, users can simply say, "SINK, open Chrome" or "SINK, search YouTube for Python tutorials." A demo video or screenshots of SINK in action will be added here. Demo highlights: Project: SINK, Local Voice AI Desktop Assistant Repository: https://github.com/still4shrishti/SINK- https://github.com/still4shrishti/SINK- Tech stack: Python, PySide6, Faster-Whisper, Qwen 3, LM Studio, PyAutoGUI, and pyttsx3. SINK combines open-source tools and locally running AI to create a voice-controlled desktop assistant. I built the assistant around practical commands, allowing users to interact with their computer through natural voice instructions instead of relying only on traditional interfaces. The project also includes productivity features such as tasks, reminders, timers, clipboard assistance, dictation, and study quizzes. Open innovation makes projects like SINK possible by allowing developers to combine existing open-source tools and open-weight AI models into something useful and personal. Running Qwen 3 locally through LM Studio gives SINK an AI brain without requiring every conversation to be sent to a paid cloud AI API. Faster-Whisper enables local speech recognition, while Python libraries provide the tools needed for desktop automation. This approach offers more control over privacy, customization, and operating costs. It also lets developers experiment, modify components, and build new capabilities without depending entirely on a single closed platform. For me, open innovation means taking powerful technologies that already exist and combining them into something people can actually use. SINK is designed to reduce the small, repetitive tasks that interrupt work and study. Instead of manually opening a browser, searching for a tutorial, switching windows, or navigating through multiple controls, users can issue a voice command and let SINK handle supported actions. For example: The real payoff is making everyday computer use feel more conversational and reducing the effort needed to complete routine tasks.