How would I go about finding sound peaks?

I am currently trying to make a sound peak visualizer and I'm struggling to make a sound peak formula. I had tried using the difference between 1 loudness to another and if the difference is big enough it will register it as one but it is too inaccurate. I am wondering if there a better possible method of finding sound peaks and when found it starts an action?

(A simple formula would be accepted)

If you don't know what sound peaks are they're the beat of a song in a nutshell. I wanna detect if a beat happens