Text to Speech for Dzongkha Language

Abstract

Text to Speech plays a vital role in imparting information to the general population who have difficulty reading text but can understand spoken language. In Bhutan, such system were not available and building a system will leverage the use of the language by different segment of people. This article describes an attempt to create a functioning model of a Text to Speech system for the Dzongkha language by creating a transcription or grapheme table for phonetic transcription from Dzongkha text to its comparable phone set. The transcription tables for consonants and vowels have been produced in such a way that they allow for improved computer compatibility. 3000 phrases were painstakingly transcribed and recorded using a single male voice. On the FESTIVAL platform, the voice synthesis was based on a statistical approach with concatenative speech creation. The model is generated using the two variants CLUSTERGEN and CLUNITS of the FESTIVAL speech tools FESTVOX where the earlier method produce more natural speech than the later for the large data set. The development of system prototype is of the first kind for the Dzongkha language in spite of attempts being made by researchers.