The Simple4All project enables the building of speech synthesis systems automatically from data with little or no expert supervision, thus creating speech synthesisers for new languages or domains in a quick, easy and pain-free way. Whilst substantial effort has been invested over the years into developing novel and progressively more natural sounding speech synthesis techniques, there has always been a high entry barrier to building a speech synthesis system in a new language, because conventional methods require expensive linguistic resources, detailed expert knowledge, and a high degree of skill in system etuningf. The result is that only a very few of the world's languages have speech synthesis systems. For those less widely-spoken languages that do have a system available, the quality of the synthetic speech is not very good, because of the lack of resources available and the lack of skills to etunef such systems. The main objectives of Simple4All are to address this imbalance by providing Open Source tools to: . enable the creation of a voice for any domain in any language, . provide an end-to-end framework in which these voices can be automatically built and improved, . produce natural and expressive speech synthesis, . provide feedback-driven learning to improve systems.
Bibliographic reference. Clark, Robert A. J. (2013): "Simple4all", In INTERSPEECH-2013, 2654-2656.