7th International Conference on Spoken Language Processing

September 16-20, 2002
Denver, Colorado, USA

A Text-to-Speech Synthesis System for Telugu

Jithendra Vepa (1), Jahnavi Ayachitam (2), K. V. K. Kalpana Reddy (3)

(1) University of Edinburgh, U.K.; (2) Jawaharlal Nehru Technological University, India; (3) University of Central Florida, USA

In this paper, a diphone based Text-to-Speech (TTS) system for the Telugu language is presented. Telugu is one of the main south-Indian languages spoken by more than 100 million people. Speech output is generated using the Festival Speech Synthesis System and the MBROLA synthesis engine. The design and collection of diphones and voice building process are described. Our text analysis module, the methods used for segment duration and generation of pitch contours are briefly discussed. Also, we present waveform generation techniques used in both MBROLA and Festival synthesis systems.


Full Paper

Bibliographic reference.  Vepa, Jithendra / Ayachitam, Jahnavi / Reddy, K. V. K. Kalpana (2002): "A text-to-speech synthesis system for telugu", In ICSLP-2002, 157-160.