J.A.R.V.I.S.

01Overview

A cross-platform AI voice assistant combining speech recognition, conversational AI and real-time interaction across desktop and web.

02Challenge

Explore a more natural way to interact with software through voice. The technical challenge is to coordinate speech recognition, AI responses and speech synthesis while keeping the interface clear about what is happening.

03Approach

Treat voice and visual feedback as one experience. An interactive, futuristic interface communicates when the assistant is ready, listening, processing or responding.

04Solution

An independent personal assistant developed by Enzo Sena in two complementary forms: a desktop application with local voice processing, and a browser-based web/PWA experience with AI integration through an API.

05Desktop application

Built with Tauri and React / Vite. Desktop voice processing uses whisper.cpp for speech-to-text and Piper for text-to-speech, within an interactive interface that makes the assistant’s state visible.

06Web / PWA experience

Accessible through the browser, with an experience adapted for mobile devices. The web version combines browser speech recognition and speech synthesis with a serverless architecture and AI integration through an API.

07Voice interaction

Speech input, processing and spoken responses form the interaction cycle. Visual states — ready, listening, processing and responding — help the user follow each stage in real time.

08AI integration

In the web version, conversational AI is integrated through an API. This connects voice interaction with AI responses within the browser-based experience.

09Technologies used

Desktop: Tauri, React, Vite, whisper.cpp and Piper. Web / PWA: browser speech recognition, speech synthesis, an AI API and a serverless architecture.