No description
  • JavaScript 35.2%
  • TypeScript 23.2%
  • CSS 22.1%
  • Python 11.2%
  • Svelte 7%
  • Other 1.3%
Find a file
G-Fourteen 17747d0364
Some checks failed
CI / build-and-test (push) Has been cancelled
Fix URL formatting in README.md
2025-11-04 15:12:49 -07:00
.claude feat: complete voice interaction migration to Svelte + TypeScript 2025-11-03 14:40:51 -08:00
.github/workflows ci: allow GitHub Pages deployment despite E2E test failures 2025-11-03 16:16:39 -08:00
AI fix: resolve all failing e2e tests 2025-11-03 12:02:02 -08:00
public feat: complete voice interaction migration to Svelte + TypeScript 2025-11-03 14:40:51 -08:00
scripts fix: attach click event listener to launch button 2025-11-03 14:06:27 -08:00
src feat: complete voice interaction migration to Svelte + TypeScript 2025-11-03 14:40:51 -08:00
tests refactor: modernize Talk-to-Unity to Vite + Svelte + TypeScript 2025-11-03 14:22:55 -08:00
.gitignore refactor: modernize Talk-to-Unity to Vite + Svelte + TypeScript 2025-11-03 14:22:55 -08:00
actionplan.md Add files via upload 2025-11-02 15:28:55 -07:00
ai-instruct.txt Add files via upload 2025-11-02 22:42:59 -07:00
BROWSER_TESTING_GUIDE.md Add files via upload 2025-11-02 20:44:02 -07:00
index.html refactor: modernize Talk-to-Unity to Vite + Svelte + TypeScript 2025-11-03 14:22:55 -08:00
index.html.legacy refactor: modernize Talk-to-Unity to Vite + Svelte + TypeScript 2025-11-03 14:22:55 -08:00
landing.js fix: refactor launch button logic for reliable navigation 2025-11-03 13:03:37 -08:00
MIGRATION.md refactor: modernize Talk-to-Unity to Vite + Svelte + TypeScript 2025-11-03 14:22:55 -08:00
package-lock.json refactor: modernize Talk-to-Unity to Vite + Svelte + TypeScript 2025-11-03 14:22:55 -08:00
package.json refactor: modernize Talk-to-Unity to Vite + Svelte + TypeScript 2025-11-03 14:22:55 -08:00
playwright-server.js Add CI workflows and stabilize image clipboard flow 2025-11-02 01:50:34 -07:00
playwright-server.mjs fix: Configure Playwright test server to properly serve from dist/ directory 2025-11-03 16:08:08 -08:00
playwright.config.mjs fix: use cross-platform environment variable syntax in Playwright config 2025-11-03 16:15:27 -08:00
README.md Fix URL formatting in README.md 2025-11-04 15:12:49 -07:00
requirements.txt Revive voice experience with dark theme 2025-11-01 09:27:48 -06:00
SPEECH_RECOGNITION_ANALYSIS.md Add files via upload 2025-11-02 20:44:02 -07:00
style.css Add files via upload 2025-11-02 20:44:02 -07:00
svelte.config.js refactor: modernize Talk-to-Unity to Vite + Svelte + TypeScript 2025-11-03 14:22:55 -08:00
tsconfig.json refactor: modernize Talk-to-Unity to Vite + Svelte + TypeScript 2025-11-03 14:22:55 -08:00
tsconfig.node.json refactor: modernize Talk-to-Unity to Vite + Svelte + TypeScript 2025-11-03 14:22:55 -08:00
vite.config.ts fix: add Vite plugin to create .nojekyll for GitHub Pages 2025-11-03 15:36:25 -08:00

DEPRICATED APPLICATION try instead: https://github.com/Unity-Lab-AI/Talking-with-Unity.git https://www.unityailab.com/Talking-with-Unity/

Talk to Unity in Plain English

Main Branch Workflow Status Pull Request Workflow Status

Talk to Unity is a single web page that acts like a friendly concierge. The landing screen double-checks that your browser has everything it needs (secure connection, microphone, speech tools). Once every light turns green, a voice assistant named Unity wakes up so you can talk out loud and hear it answer back.

What you need

  • A recent version of Chrome, Edge, or Safari. (Firefox still lacks the speech tools we use.)
  • A secure address — either https:// on the web or http://localhost while testing.
  • Speakers or headphones so you can hear Unity talk.
  • A microphone and permission to use it.

If any of these are missing, the landing page will highlight what to fix and share a short tip.

Fastest way to try it

  1. Download the project or clone it with Git.
    git clone https://github.com/Unity-Lab-AI/Talk-to-Unity.git
    cd Talk-to-Unity
    
  2. Start a basic web server. Python is an easy option:
    python -m http.server 8000
    
  3. Open http://localhost:8000 in a supported browser and load index.html.
  4. Watch the landing page run its checks. When everything shows Ready, press the button to enter the AI lab and start talking to Unity.

Understanding the landing page

  • System lights: Secure context, speech recognition, speech playback, and microphone access each get a color. Green means you are good to go. Amber means you will see clear instructions to fix the problem.
  • Status message: The box at the top always tells you what Unity is waiting on. When all lights are green it invites you into the lab.
  • Help links: Quick links back to the Unity AI Lab website and this repository sit near the top so you can jump out for more info.

Using the voice assistant

  1. Enter the lab once the landing checks pass.
  2. Click Unmute microphone to let Unity listen.
  3. Speak naturally. Unity will transcribe, respond with text, and also read its answer aloud.
  4. Use the on-screen controls to mute, stop, or reset the conversation if you want to start over.

All of this runs in the browser — no extra servers or databases are required.

Customizing the experience

  • Prompt and personality: Edit ai-instruct.txt to change Unitys default instructions.
  • Landing behavior: Adjust the readiness checks or UI copy inside landing.js.
  • Voice interface: Modify AI/app.js for how Unity listens, thinks, and speaks.
  • Styling: Tweak the look across the landing page and chat area in style.css and AI/style.css.

File map

File Purpose
index.html Loads the landing page, pulls in the styles, and boots both scripts.
landing.js Handles the pre-flight checks and controls the landing layout.
style.css Shared styling for the landing screen.
AI/app.js Runs the live voice conversation.
AI/style.css Styles for the in-lab chat interface.
ai-instruct.txt Text prompt that shapes how Unity responds.

Need more help?

Share feedback, swap in your own prompt, or dress up the visuals — just keep it simple and have fun chatting with Unity.