Limited Domain Synthesis
Alan W. Black, Kevin Lenzo · 2000
This work presents a reliable and efficient method for building limited domain speech synthesis voices. By constructing databases close to the targeted domain of the speech application, unit selection synthesis techniques can be used to reliably give very high quality synthesis within domain. In addition to a high quality result we include the techniques and processes required to build such voices often allowing new voices in limited but quite complex domains such as dialog systems to be created in under a week. The full tools, documentation examples etc are available for free at http://festvox.org. 1. INTRODUCTION With the recent increase in demand for speech applications, it has become obvious that current general speech synthesis technology is not at a quality that users accept. Many speech applications still use fixed, fully pre-recorded prompts rather than standard TTS (text-to-speech) systems to generate their speech output, because the quality of standard TTS systems is not pe...