One important concept that applies to all forms of dialogue production is that of ducking. Ducking is the process of fading down music levels when someone begins speaking. In other words, the music “ducks” under the dialogue. Simply begin a fade slightly before the voice begins speaking, and then complete the fade somewhere around the first syllable of speech. You want to make sure the change in volume isn’t awkwardly sudden, but also that the music doesn’t affect the intelligibility of the first spoken word.