Spread gives an offset value which is different for each of the voices which are playing the same midi note (Unison). When only one voice is playing a note, the value is zero. The offsets are grouped around zero and are multiples of the value at the module's input.
If someone can give ideas of what it is used for that would be most welcome.