Connor Copeland Undergraduate Dissertation 2016/17

Controlling a Speech Synthesiser through Gesture

Supervised by R.K.Moore

Abstract

Significant progress has been made in the field of speech generation in the past two decades with many us experiencing artificial speech regularly, if not, every day. Whilst being depicted as robotic in films, speech synthesisers are, in reality, capable of producing speech almost indistinguishable from real human speech.

Whilst there are a multitude of ways artificial speech can be generated, many are reliant on text-based input therefore the ability to generate speech that conveys the expression and emotional output that real speech can produce is limited.

The aim of this project, therefore, is to produce an interactive system that enables people to synthetically generate speech in real time, using analogue movements, allowing expressive vocal behaviour to be communicated. The system would be suitable for vocally disabled users and as a novel way for expressing speech that may be of interest to artists.