One group of prospective mothers, expose to A only prenatally.

The other group, expose to B only.

After birth, expose both groups to both A and B and determine preference.

The trick is having a repeatable way to determine preference. eg a smile or frown appearing within ‘x’ seconds of starting the story or piece of music.

You could probably also have a third group, which is not exposed to anything prenatally. But that is really a different experiment (ie does prenatal exposure to ANY songs/stories change how a newborn baby reacts to songs/stories in general).

So I would just stick with the two groups. Simpler.

key issues here are a group of pregnant woman to test.

a control group of pregnant women as a control group.

record all stories or music the women are exposed to in normal life , or lock them up so there is none

expose the test group to selected stories and music , and maybe use one song thats really odd , like accordian music.

then after birth test the childs preferences and compare the test and control groups.