Direct Text to Speech Translation System using Acoustic Units