Development of Konkani TTS system using concatenative synthesis

Abstract

Text to Speech (TTS) Synthesizer is an application that converts text to speech. Research on TTS technology has received attention among researchers in Indian languages resulting in its use in many applications e.g. to read out text for visually or vocally impaired people. However, Konkani, the official language of Goa in India has gained lesser attention as compared to other Indian languages in this context. Our previous work, focused on speech synthesis for Konkani language using eSpeak tool. eSpeak uses formant synthesis and the synthetic speech generated is considered to be unnatural (robotic) in the Indian Context. This paper describes our recent investigation on the use of acoustic units by concatenative speech synthesis and discusses the issues of using different types units: words, diphones, and phonemes as a database. A GUI has been developed for converting Konkani text into speech. The use of different unit size for synthesis show better performance of word synthesizer over other synthesizers. The subjective listening test shows a performance improvement in reading word, phrase and sentence.