It is a bitwise NOT.
Most common use I’ve seen is a double bitwise NOT, for removing the decimal part of a number, e.g:
var a = 1.2;
~~a; // 1
Why not use Math.floor
? The trivial reason is that it is faster and uses fewer bytes. The more important reason depends on how you want to treat negative numbers. Consider:
var a = -1.2;
Math.floor(a); // -2
~~a; // -1
So, use Math.floor
for rounding down, use ~~
for chopping off (not a technical term).