Jarvis: A Browser-Based Bilingual AI Voice Assistant with Multi-Model LLM Selection and HUD-Style Interface Design

Most consumer AI voice assistants are delivered as native mobile or desktop applications requiring installation, account registration, and dependency on a single fixed language model controlled by the platform provider. This paper presents Jarvis, a browser-based AI voice assistant requiring no installation, built entirely with HTML, CSS, and JavaScript on the client side, paired with the Groq API for language model inference. Jarvis supports bilingual interaction in English and Malayalam, offers users a choice among multiple free large language models, and presents a science-fiction-inspired heads-up display (HUD) interface featuring an animated reactor visualization, real-time voice waveform visualization, and a scanning-line motion effect. The assistant maintains conversational memory within a session and exposes a user-supplied API key model, allowing each user to authenticate with their own free Groq API credentials rather than relying on a shared backend service.