An overview of text-to-speech synthesis

Abstract

Summary form only given. The article gives an overview of text-to-speech (TTS) technology and a description of some issues of potential interest to speech coding experts. After motivation for the use of TTS technology, it describes the general architecture of a text-to-speech system with particular emphasis on the speech synthesis component. Both formant synthesis and concatenative synthesis are presented, offering different degrees of flexibility and quality. Several well-known speech coding techniques (including LPC vocoders, waveform interpolation, harmonic coding, and layered coding) have been used in speech synthesis. It explains how they have been applied, and the advantages and limitations of those techniques when used in speech synthesis. The main goal is to increase cooperation between the speech coding community and the TTS community, and in particular to motivate the need for speech coding algorithms that meet the requirements of the next generation speech synthesis technology.