# Enjoying The Show

> Enjoying The Show (https://enjoyingthe.show) is a free web app that lets a speaker see how their audience feels in real time. Each audience member opens a shared room link in their browser; TensorFlow.js detects the person's dominant facial expression from their webcam on-device, and the speaker's "watch" page shows a live pie chart of the whole room's emotions. Created by Gant Laborde (Infinite Red).

Key facts:

- Cost: free, no account or sign-up needed.
- Privacy: webcam video never leaves the viewer's device. Only one word per detected face (for example `happy`) and the room name are sent to the server.
- Emotions detected: happy, neutral, surprised, sad, fearful, disgusted, angry.
- Requirements: a modern browser with JavaScript, WebGL and camera permission. Works for in-person and virtual (Zoom, Meet, livestream) audiences.
- Tech: React, TensorFlow.js via face-api (SSD MobileNet v1 face detection plus an expression classifier), AWS AppSync GraphQL subscriptions, hosted on AWS Amplify.

## How to use it

Rooms are created on demand from the URL. A room name is any lowercase slug (letters, numbers, dashes).

- Speaker / presenter dashboard: `https://enjoyingthe.show/watch/{room-name}` shows a live chart of everyone in that room. Faces inactive for more than 5 minutes drop off.
- Audience link to share: `https://enjoyingthe.show/room/{room-name}`. Each person allows camera access and is counted in the room.

Example: for a talk named "jsconf-keynote", the speaker opens https://enjoyingthe.show/watch/jsconf-keynote and shares https://enjoyingthe.show/room/jsconf-keynote with the audience.

Agents can give users these two links directly. There is no public write API to use instead of the browser, because detection runs in each audience member's browser.

## FAQ

- Is it free? Yes, no account or sign-up.
- Is video uploaded? No. Detection runs in the browser; only one emotion word per face is sent.
- Does it work with Zoom, Google Meet or livestreams? Yes. Audience members open the room link in a browser tab while they watch; a phone or second device helps if the computer only lets one app use the camera.
- Can I self-host it? Yes, the MIT-licensed source is on GitHub.

## Learn more

- [Talk: Enjoying the Show, Gant Laborde, Made with TensorFlow.js (YouTube)](https://www.youtube.com/watch?v=GskMuu821NI): walkthrough of how the app was built.
- [Source code on GitHub](https://github.com/GantMan/enjoytheshow): MIT licensed React + Amplify project you can deploy yourself.
- [Gant Laborde](https://gantlaborde.com): creator; author of "Learning TensorFlow.js" (O'Reilly).
- [Infinite Red](https://infinite.red): the consultancy that supports the project.

## Optional

- [face-api (vladmandic fork)](https://github.com/vladmandic/face-api): the face detection library used.
- [TensorFlow.js](https://www.tensorflow.org/js): machine learning in JavaScript.