← Back to projects

EzSpeak

A Chrome extension for real-time captions, translation, and AI voice.

2nd Place — Google AI for Social Good Challenge, ShellHacks 2025
View code

Demo

Overview

EzSpeak breaks down language barriers in live meetings and streams. It captions what's being said, translates it, and can speak it back in an AI voice in the language you choose, all in real time. We built it in 36 hours at ShellHacks 2025, where it placed 2nd in Google's AI for Social Good Challenge.

What I worked on

  • Led the audio pipeline and Azure AI Speech integration on a 3-person team.
  • Captured tab audio using Chrome MV3 and an AudioWorklet, then ran it through the Azure Speech SDK for speech-to-text, translation, and text-to-speech.
  • Kept transcription and translation under 2 seconds of latency, with an AI voice in the chosen language.