When a character in your story tells a story, you have to decide which voice controls dialogue in that story. Example: A young boy is speaking, telling someone else what his father said to him.

If you narrate the father's voice as an adult voice, it will be jarring. What you most likely want is the son's imitation of his father's voice. It is short enough.

On the other hand, if the story in a story is longer, then in most cases you should use the voices of the characters in the story in the story (use the father's voice, not the son imitating the father's voice). How long does it have to be?

The reason for shifting or not shifting the voices involves the choice of perspective that you are giving to this part of the book. If it is important to say in the present moment (when the boy is talking), there is no shift. But if the story in the story has to take center stage, then you shift to the past and use the father's voice to create a "present in the past".