Approach

Matrubhasha fundamentally provides a Rule-Based TTS system for Indian languages. For Speech Synthesis, Matrubhasha uses Mbrola - a speech synthesizer based on concatenation of diphones, which is developed by the TCTS lab of 'Faculte Polytechnique de mons' (Belgium) and provided free for non-commercial applications. It takes a list of phonemes as input, together with prosodic information (duration of phonemes and a piecewise linear description of pitch), and produces speech samples on 16 bits (linear), at the sampling frequency of the diphone database used.
Matrubhasha aims at synthesizing speech from a given Unicode input, by converting it into a list of phonemes, together with prosodic, information in the format required by Mbrola speech synthesizer. Matrubhasha also aims at providing plug-ins with TTS functionalities for commonly used software applications like Office Applications and Internet Browsers. Matrubhasha is also working towards emotional speech output, which sounds more natural.

Architecture

The architecture of Matrubhasha with it's subsystems is as shown below:




Mr. Raman and the Matrubhasha Team

More