University Court of the University of Edinburgh

Published purpose

This proposed research contributes to the wider Small Teams project, which aims to develop text-to-speech synthesis voices to support learning of Indigenous languages in Canada. It will be made possible to hear words and phrases pronounced authentically and in a variety of voices on demand, so enriching language learners’ self-study. Educators will have a tool to author speech passages such as stories or dialogues for students. Furthermore, this collaboration aims to adapt synthesis to better suit language learning in two ways: i) to enable a user to hear synthesized words in a carefully spoken form, so helping beginners learn pronunciation and vocabulary; ii) allow to influence and manipulate the synthesis in a maximally intuitive way, so to ensure speech samples sound exactly as a educator-user intends. Finally, the project will also aim to improve ways both of evaluating synthetic speech automatically and to streamline listening tests to reduce demands on human listeners.