Marathi Language Speech Synthesizer Using Concatenative Synthesis Strategy (Spoken in Maharashtra, India)

Abstract

in this paper, we present the concatenative text-to-speech system and discuss the issues relevant to the development of a Marathi speech synthesizer using different choice of units: words, phonemes as a database. Quality of the synthesizer with different unit size indicates that the word synthesizer performs better than the phoneme synthesizer. The most important qualities of a speech synthesis system are naturalness and intelligibility. We synthesize the Marathi text and perform the subjective evaluations of the synthesized speech. As a result, (1) 81% of speech synthesized by the proposed method was preferred to that by the conventional method, (2) The error rate of TTS synthesizer is around 8.22%, (3) Speech synthesis runtime was reduced for proposed method. The results show the effectiveness of the proposed method.