In this repo there are a few basic css files you can use as starting points.
PNGTube uses the name of each image for its id (the # items in css) and applies the 'active' class (the . items in css) when audio is detected.
You can replicate this behavior in the test links by clicking on the window to toggle the active class.
basic.css test All other files are based on this one, which changes the opacity of the 'active' and 'idle' images based on voice activity
brightness.css test This file lowers the brightness of the idle image
bounce.css test This file adds a bounce animation when speaking starts
blink.css test This file adds a blink animation, and requires active-blink and idle-blink images.
reactive.ss test This file simply combines the brightness and bounce files to replicate the Reactive by FugiTech style.
hide.css test This file hides all images below the viewport during silence