Assigning Intonational Features in Synthesized Spoken Directions