Is there a valid reason why you don’t want to use a blendspace? With four different speed values and three different positions, a 2d blendspace would be perfect for this. Having separate states in a state machine will be a mess since you’ll have to interconnect almost all of them back and forth, something like this:
And setting all the conditions will be hell. Whereas in a 2d blendspace you can have speed as one axis, and position as another (let’s say, 0 for prone, 1 for crouch, 2 for upright), and all animations will be blended between automatically.