Speech Synthesis Using Perceptually Motivated Features.

摘要

The trajectory of this project parallels, in certain ways, the changing dynamics of speech and neuroscience research. When the project began in 2007, many aspects of these fields were formulated in concepts and methods originating more than 50 years ago. By the project's conclusion, in May 2011, the focus in both fields had shifted to statistical (often Bayesian) approaches, with a clear recognition that the classical models require serious revision. Whether the Bayesian framework turns out to be the 'best' one is uncertain. Despite its limitations (narrow perspective, lack of explanatory insight), it represents a significant improvement over traditional, quasi-deterministic approaches that dominated speech and brain research over much of the 20th century.