Statistical TTS systems that directly predict the speech waveform have recently reported improvements in synthesis quality. This investigation evaluates Amazon's statistical speech waveform synthesis (SSWS) system. An in-depth evaluation of SSWS is conducted across a number of domains to better understand the consistency in quality. The results of this evaluation are validated by repeating the procedure on a separate group of testers. Finally, an analysis of the nature of speech errors of SSWS compared to hybrid unit selection synthesis is conducted to identify the strengths and weaknesses of SSWS. Having a deeper insight into SSWS allows us to better define the focus of future work to improve this new technology.

No abstract

BT is currently developing a low computation, real time, talking head as an adjunct to the Laureate text-to-speech systemrl].Research into the development of a talking head may be divided into two components; image generation, and face and head movement control. This paper concentrates on the last of the two.A significant aspect of this work is research into methods of generating convincing mouth shapes when the head is tallcing. The paper describes a real time method of visual speech generation, which takes into consideration major coarticulation effects. It provides a detailed description of the generation process and compares this with a method of visual speech generation proposed by Cohen and Massaro[2].

Phonetic segmentation is an important step in the development of a concatenative TTS voice. This paper introduces a segmentation process consisting of two phases. First, forced alignment is performed using an HMM-GMM model. The resulting segmentation is then locally refined using an SVM based boundary model. Both the models are derived from multi-speaker data using a speaker adaptive training procedure. Evaluation results are obtained on the TIMIT corpus and on a proprietary single-speaker TTS corpus.

No abstract