Building a Voice-Enabled AI Assistant With Javascript
summary
Builds a Jarvis-style assistant using Web APIs and langchain, with live coding showing actual code: Chrome-only speech recognition API (webkit prefix), speech synthesis, fetch to OpenAI GPT‑3.5‑turbo, and a simple TypeScript/Vite setup. Shows the loop that restarts the mic after speaking, explains the need for a user click, uses OS voices only, requires an API key, and keeps conversation state in an array. Result is a working demo in under fifty lines that runs in the browser and speaks back.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:57Planning with browser APIs
- 4:02Speech-to-text with Web Speech API
- 5:48Connecting to GPT-3.5 Turbo
- 10:32Text-to-speech with Speech Synthesis API
- 13:43Continuous conversation loop
- 16:09Final code and limitations
- 18:58Closing thoughts
related talks
38mIntegrating AI and Browser API Voice Recognition on Web App
01h 26mBuilding voice agents with OpenAI
51mBuilding an AI assistant that makes phone calls
30mLook ma, no hands! Multimodal AI-Agents in the browser
29mFinding Your Voice: Building Screenless Interfaces with Node.js
38mConversational UIs: Talking to Siri, Alexa, and Your Web Browser
49mHello JARVIS - Building Voice Interfaces for Your LLMS
27mWeb AI: Magic or Myth