Generate audio from text using a voice synthesis model