HMM Based Text-to-Speech Synthesis for Telugu

Gugulothu, Narendhar (2016) HMM Based Text-to-Speech Synthesis for Telugu. Masters thesis, Indian Institute of Technology Hyderabad.

Preview

Text
EE14MTECH11004.pdf - Submitted Version
Download (1MB) | Preview

Abstract

This thesis describes a novel approach to build a general purpose working Telugu text-to- speech synthesis system (TTS) based on hidden Markov model (HMM) which is reasonably intelligible, natural sounding and exible. There have been several attempts proposed to use HMM for constructing TTS systems. Most of such systems are based on waveform concatenation techniques. To fully convey information present in speech signals, text-to-speech synthesis systems are required to have an ability to generate natural sounding speech with arbitrary speakers individualities and emotions (e.g., anger, sadness, joy). To represent all these factors the Mel- cepstral coefficients are extracted as spectral parameters. Excitation parameters are extracted using fundamental frequency(F0).

[error in script]

IITH Creators: