A link to this repos is a good start.

And reorganize the source as a module.

It will work better if listening to input changes and parsing the whole command each time.

Ask permission to access audio input (e.g. mic) and start changing to the next video on each new BPM. Needs to use Web Audio API to detect BPM.

Can be improved later to spectrum analysis, amplitude level or volume.

Maybe displaying it as texture and processing it with WebGL shaders.

Useful to make it possible to turn it on/off.