The problem SneakyLeaks.js solves

😖 The Problem

The reason for it being loved by people is the rawness & interactivity. People feel more connected while watching streams. BUT, it all comes with its own downsides. I've seen many streamers messing up live by leaking phone numbers, sensitive information, or private conversations unknowingly. After all, the lack of control & rawness isn't always good for the streamer.

🚀 The Solution

Hence, in order to avoid these Sneaky Leaks from messing up live streams & even entire careers at times, we have created a Machine Learning based feature which detects what you're doing and toggles the audio or video accordingly. Currently, we've trained it to perform the following tasks:

• Checks whether you're wearing headphones & mute your microphone accordingly.
• Turns the video off when you're not in the frame to avoid your background conversations being overheard.
• Mutes the microphone when it sees that you're on a phone call to prevent sensitive conversations from being leaked.

Challenges I ran into

The major challenges I ran into was figuring out a way to make the output viewable anyhow & preferably on the web. Thankfully I found p5 JavaScript which turned out to be pretty efficient.